[SI-LIST] : IMAPS conf on high speed digital interconnects

About this list Date view Thread view Subject view Author view

From: Eric Bogatin ([email protected])
Date: Thu Jan 04 2001 - 13:44:44 PST


Hi folks-

Just a last minute reminder about the IMAPS Advanced Technology
Workshop next week on High Speed Digital Interconnects. You can find
more information at the url: http://www.imaps.org/hispeed2001prog.htm

see ya there!

--eric

**************************************
Eric Bogatin
CTO, Giga Test Labs
v: 913-393-1305
f: 913-393-1306
e: [email protected]
web: www.gigatest.com
**************************************

-----Original Message-----
From: [email protected]
[mailto:[email protected]]
Sent: Wednesday, January 03, 2001 7:00 PM
To: [email protected]
Subject: si-list-digest V1 #377

si-list-digest Wednesday, January 3 2001 Volume 01 :
Number 377

In this issue:

1.) RE: [SI-LIST] : SMA test connector PCB layout
        "Larry Miller" <[email protected]>

2.) [SI-LIST] : simulation software
        "Sean Murray" <[email protected]>

3.) [SI-LIST] : UltraCADs ESR Calculator *UPGRADED* (Resend)
        Doug Brooks <[email protected]>

4.) [SI-LIST] : Re: si-list-digest V1 #376
        "Ken Taylor" <[email protected]>

5.) [SI-LIST] : Differential Impedance Effects on Diff. Return Current
        "Dill, Franz @ Celerity" <[email protected]>

6.) [SI-LIST] : RE: CDL file format
        Doug Hopperstad <[email protected]>

7.) RE: [SI-LIST] : Differential Impedance Effects on Diff. Return Cu
rrent
        "Ingraham, Andrew" <[email protected]>

8.) Re: [SI-LIST] : Differential Impedance Effects on Diff. Return
Current
        Martyn Gaudion <[email protected]>

9.) Re: [SI-LIST] : Differential Impedance Effects on Diff. Return
Current
        Doug Brooks <[email protected]>

10.) RE: [SI-LIST] : Differential Impedance Effects on Diff. Return Cu
rrent
        "Chan, Michael" <[email protected]>

11.) RE: [SI-LIST] : Differential Impedance Effects on Diff. Return Cu
rrent
        "Dill, Franz @ Celerity" <[email protected]>

12.) Re: [SI-LIST] : Differential Impedance Effects on Diff. Return
Current
        Michael Nudelman <[email protected]>

13.) RE: [SI-LIST] : Differential Impedance Effects on Diff. Return Cu
rrent
        "Dunbar, Tony" <[email protected]>

14.) Re: [SI-LIST] : RE: CDL file format
        Alan Zhu <[email protected]>

----------------------------------------------------------------------

Date: Wed, 3 Jan 2001 11:33:41 -0800
From: "Larry Miller" <[email protected]>
Subject: RE: [SI-LIST] : SMA test connector PCB layout

This is a multi-part message in MIME format.

- ------=_NextPart_000_000E_01C07579.06C6D340
Content-Type: text/plain;
        charset="us-ascii"
Content-Transfer-Encoding: 7bit

Thanks! That's exactly what I ordered.

I agree about "caring about" discontinuities on TDR measurements.
However,
they are not so hidden from E-Z view in TDR as in a frequency-domain
measurement.

I've had a number of 1 Gb SERDES manufacturers' evaluation boards with
conventional SMA connectors with no problem (as well as the end-fire
setups)
but these were comparatively short trace lengths, etc., etc.

Larry Miller
  -----Original Message-----
  From: [email protected]
[mailto:[email protected]]On Behalf Of Bob Lewandowski
  Sent: Wednesday, January 03, 2001 10:41 AM
  To: Larry Miller
  Cc: Si-List (E-mail)
  Subject: Re: [SI-LIST] : SMA test connector PCB layout

  Johnson Components has an "end launch" SMA that fits into a slot in
the
board edge, with a pin to contact a top surface microstrip trace.
It's
application is independent of board thickness. Johnson P/N
142-0721-88x.
Their web site is: http://www.johnsoncomp.com/sma.htm. A front view
picture is shown as "End Launch". The pin diameter is 30 mils, so
some
adjustment of the launch must be made for trace widths less than 30
mils.
Connection from the body to ground is also a significant issue with
this
type of connector. An edge plated slot that ties directly to the
ground
layers is best, both electrically and mechanically.
  Your other choice, the 4 ground pin through hole type is difficult
to
match to FR-4 or similar Er materials. The best possible application
is to
a back side microstrip with minimum center conductor hole and pad
diameters.
No pads on inner layers. Clear ground pads on all layers back as far
as
possible, and drill holes after plating between the center pin and the
corner pins to reduce capacitance. With careful layout you can get a
reflection coefficient of better than 10%.

  Incidentally, you should care about bad launches with a TDR because
a
large launch discontinuity masks (attenuates) downstream reflections.

  ---Bob Lewandowski
      Vixel Corp.

  Larry Miller wrote:

    I need to make some measurements on PC boards up to 5 GHz using a
network
    analyzer. In the past I didn't care that much because I was doing
TDR
    measurements.
    Unfortunately, the boards I will be looking at are too thick to
use the
"SMA
    edge connector" trick.

    Does anyone have a pointer to a good PCB footprint that will not
have
    impedance discontinuities. The connector I want to mount is the
familiar
    center pin + 4 ground pins vertical SMA style.

    Thanks in advance for any help,

    Larry Miller

    **** To unsubscribe from si-list or si-list-digest: send e-mail to
    [email protected]. In the BODY of message put:
UNSUBSCRIBE
    si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
    si-list archives are accessible at http://www.qsl.net/wb6tpu
    ****

- ------=_NextPart_000_000E_01C07579.06C6D340
Content-Type: text/html;
        charset="us-ascii"
Content-Transfer-Encoding: quoted-printable

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<HTML><HEAD>
<META HTTP-EQUIV=3D"Content-Type" CONTENT=3D"text/html; =
charset=3Dus-ascii">

<META content=3D"MSHTML 5.00.3103.1000" name=3DGENERATOR></HEAD>
<BODY>
<DIV><FONT color=3D#0000ff face=3DArial size=3D2><SPAN=20
class=3D226322619-03012001>Thanks! That's exactly what I=20
ordered.</SPAN></FONT></DIV>
<DIV><FONT color=3D#0000ff face=3DArial size=3D2><SPAN=20
class=3D226322619-03012001></SPAN></FONT>&nbsp;</DIV>
<DIV><FONT color=3D#0000ff face=3DArial size=3D2><SPAN =
class=3D226322619-03012001>I=20
agree about "caring about" discontinuities on TDR measurements.
However, =
they=20
are not so hidden from E-Z view in TDR as in a frequency-domain=20
measurement.</SPAN></FONT></DIV>
<DIV><FONT color=3D#0000ff face=3DArial size=3D2><SPAN=20
class=3D226322619-03012001></SPAN></FONT>&nbsp;</DIV>
<DIV><FONT color=3D#0000ff face=3DArial size=3D2><SPAN =
class=3D226322619-03012001>I've=20
had a number of 1 Gb SERDES manufacturers' evaluation boards with =
conventional=20
SMA connectors with no problem (as well as the end-fire setups) but =
these were=20
comparatively short trace lengths, etc., etc.</SPAN></FONT></DIV>
<DIV><FONT color=3D#0000ff face=3DArial size=3D2><SPAN=20
class=3D226322619-03012001></SPAN></FONT>&nbsp;</DIV>
<DIV><FONT color=3D#0000ff face=3DArial size=3D2><SPAN =
class=3D226322619-03012001>Larry=20
Miller</SPAN></FONT></DIV>
<BLOCKQUOTE style=3D"MARGIN-RIGHT: 0px">
  <DIV align=3Dleft class=3DOutlookMessageHeader dir=3Dltr><FONT =
face=3DTahoma=20
  size=3D2>-----Original Message-----<BR><B>From:</B>=20
  [email protected] =
[mailto:[email protected]]<B>On=20
  Behalf Of </B>Bob Lewandowski<BR><B>Sent:</B> Wednesday, January 03,
=
2001=20
  10:41 AM<BR><B>To:</B> Larry Miller<BR><B>Cc:</B> Si-List=20
  (E-mail)<BR><B>Subject:</B> Re: [SI-LIST] : SMA test connector
PCB=20
  layout<BR><BR></DIV></FONT>Johnson Components has an "end launch"
SMA =
that=20
  fits into a slot in the board edge, with a pin to contact a top =
surface=20
  microstrip trace.&nbsp; It's application is independent of board=20
  thickness.&nbsp; Johnson P/N 142-0721-88x.&nbsp; Their web site =
is:&nbsp; <A=20
  =
href=3D"http://www.johnsoncomp.com/sma.htm">http://www.johnsoncomp.com
/sm=
a.htm.</A>&nbsp;&nbsp;=20
  A front view picture is shown as "End Launch".&nbsp; The pin
diameter =
is 30=20
  mils, so some adjustment of the launch must be made for trace widths
=
less than=20
  30 mils.&nbsp; Connection from the body to ground is also a =
significant issue=20
  with this type of connector.&nbsp; An edge plated slot that ties =
directly to=20
  the ground layers is best, both electrically and mechanically.=20
  <P>Your other choice, the 4 ground pin through hole type is
difficult =
to match=20
  to FR-4 or similar Er materials.&nbsp; The best possible application
=
is to a=20
  back side microstrip with minimum center conductor hole and pad=20
  diameters.&nbsp; No pads on inner layers.&nbsp; Clear ground pads on
=
all=20
  layers back as far as possible, and drill holes after plating
between =
the=20
  center pin and the corner pins to reduce capacitance.&nbsp; With =
careful=20
  layout you can get a reflection coefficient of better than 10%.=20
  <P>Incidentally, you should care about bad launches with a TDR
because =
a large=20
  launch discontinuity masks (attenuates) downstream reflections.=20
  <P>---Bob Lewandowski <BR>&nbsp;&nbsp;&nbsp; Vixel Corp.=20
  <P>Larry Miller wrote:=20
  <BLOCKQUOTE TYPE=3D"CITE">I need to make some measurements on PC =
boards up to=20
    5 GHz using a network <BR>analyzer. In the past I didn't care that
=
much=20
    because I was doing TDR <BR>measurements.=20
    <P>Unfortunately, the boards I will be looking at are too thick to
=
use the=20
    "SMA <BR>edge connector" trick.=20
    <P>Does anyone have a pointer to a good PCB footprint that will
not =
have=20
    <BR>impedance discontinuities. The connector I want to mount is
the =
familiar=20
    <BR>center pin + 4 ground pins vertical SMA style.=20
    <P>Thanks in advance for any help,=20
    <P>Larry Miller=20
    <P>**** To unsubscribe from si-list or si-list-digest: send e-mail
=
to=20
    <BR>[email protected]. In the BODY of message put: =
UNSUBSCRIBE=20
    <BR>si-list or UNSUBSCRIBE si-list-digest, for more help, put
HELP.=20
    <BR>si-list archives are accessible at&nbsp; <A=20

href=3D"http://www.qsl.net/wb6tpu">http://www.qsl.net/wb6tpu>=20
  <BR>****</P></BLOCKQUOTE></BLOCKQUOTE></BODY></HTML>

- ------=_NextPart_000_000E_01C07579.06C6D340--

**** To unsubscribe from si-list or si-list-digest: send e-mail to
[email protected]. In the BODY of message put: UNSUBSCRIBE
si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
si-list archives are accessible at
http://www.qsl.net/wb6tpu
****

------------------------------

Date: Wed, 3 Jan 2001 13:24:01 -0700
From: "Sean Murray" <[email protected]>
Subject: [SI-LIST] : simulation software

Anyone have any experience with either Speed2000 from sigrity or
Eagleware
RF? It appears the eagleware is a standalone system for doing
circuits, any
truth to that?

Sean Murray
General Manager

M&M Specialties, Inc.
1236 W. Southern Ave. #106
Tempe, AZ 85282
1-800-892-8760 x105

www.mmspec.com

**** To unsubscribe from si-list or si-list-digest: send e-mail to
[email protected]. In the BODY of message put: UNSUBSCRIBE
si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
si-list archives are accessible at http://www.qsl.net/wb6tpu
****

------------------------------

Date: Wed, 03 Jan 2001 11:47:33 -0800
From: Doug Brooks <[email protected]>
Subject: [SI-LIST] : UltraCADs ESR Calculator *UPGRADED* (Resend)

I apologize for the resend, but I got back about 10,000 "out of
Office"
responses last week. Bad week to send an announcement.

Besides, since then we made one minor improvement and now 3.01 is the
current version!
**********************************************************************
******
***********

At popular demand, UltraCAD's popular ESR calculator has been upgraded
to
Version 3.01

This is a FREE upgrade! All previous licenses will work with this new
version. Obtain the upgraded version from our web site at
http://www.ultracad.com

Changes (all of which have been requested by users) include:

*****************

This version is a Windows 9x Version and therefore recognizes long
filenames. It also is a little better than Version 2 at multitasking.

It is compatible with previous licenses. Previous input files need one
minor modification to be compatible with this version (the addition of
a
single line containing the string "startdata"). Refer to the updated
help
file for details

Comments may be entered and edited on the input setup screen, the
output
screen, and can be included in the input file. There is almost no
theoretical limit to the length of comment text, but there is a
practical
limit to what can be displayed on the screens.

The user can select either a white graphical background (for better
printing) or a gray background (for easier viewing). (Unfortunately,
you
can't select both, and you can't change a setting while a graph exists
on
an output form.)(3.01 adds a check-mark here to show current setting.)

There is improved output graphical information, now presented on two
screens (for users who like to use lower screen resolutions.)

There is provision to display major grids on the graph.

There is provision for printed output of the graph and the graphical
data.

*********************

Happy New Year to all from your friends at UltraCAD Design, Inc.

Doug Brooks
.
************************************************************
Doug Brooks' book "Electrical Engineering for the Non-Degreed
Engineer" is now available. See our web site for details.
.
Doug Brooks, President [email protected]
UltraCAD Design, Inc. http://www.ultracad.com

**** To unsubscribe from si-list or si-list-digest: send e-mail to
[email protected]. In the BODY of message put: UNSUBSCRIBE
si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
si-list archives are accessible at http://www.qsl.net/wb6tpu
****

------------------------------

Date: Wed, 3 Jan 2001 14:09:56 -0800
From: "Ken Taylor" <[email protected]>
Subject: [SI-LIST] : Re: si-list-digest V1 #376

Chris, what you propose is very doable. At some point you have to
trade
theory and purity for connectivity, and both Tek and HP (now Agilent)
as
well as others long past like Philips and Dolch etc. have done it for
years
with various woven ribbon configurations ending in pairs.

Regards, Ken.

Ken Taylor
Polar Instruments, Portland, OR, USA.
+1 503 356-9001
[email protected]

> Date: Tue, 2 Jan 2001 20:47:07 -0500
> From: "Christopher R. Johnson" <[email protected]>
> Subject: [SI-LIST] : Resistive probe and twisted pair revisited
>
> This is a multi-part message in MIME format.
>
> - ------=_NextPart_000_0104_01C074FD.2B907810
> Content-Type: text/plain;
> charset="iso-8859-1"
> Content-Transfer-Encoding: quoted-printable
>
> It seems that the biggest down side to using twisted pair for the =
> resistive probe is the lack of the coaxial shield for noise
reduction =
> and impedance stabilization.
>
> The actual application is for a very low cost relatively high =
> performance logic analyzer (500 MHz sample rate), so there must be
16 =
> probes.

**** To unsubscribe from si-list or si-list-digest: send e-mail to
[email protected]. In the BODY of message put: UNSUBSCRIBE
si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
si-list archives are accessible at http://www.qsl.net/wb6tpu
****

------------------------------

Date: Wed, 3 Jan 2001 14:20:09 -0800
From: "Dill, Franz @ Celerity" <[email protected]>
Subject: [SI-LIST] : Differential Impedance Effects on Diff. Return
Current

All,

Please excuse my 'newbie-like' questions, my inclusion in this mailing
list
is more for curiosity and personal advancement/understanding than as a
profession.

First, is this statement valid?
- - In a differential pair, one 'leg' of a signal's return current
path is
through the complementary 'leg' of a differential pair and not through
the
ground or power planes (Assuming equal trace lengths, Zo=50
single-ended,
Zo=100 diff. impedance - using ECL logic as an example).

Now, assuming the above statement is true:
If the differential impedance is NOT 100 Ohms (Differential traces NOT
routed differentially) how does this effect the return current path?
Does
the return current begin to flow through the ground and power planes
rather
than through the differential pair?

Thanks SI gurus!

Franz.

**** To unsubscribe from si-list or si-list-digest: send e-mail to
[email protected]. In the BODY of message put: UNSUBSCRIBE
si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
si-list archives are accessible at http://www.qsl.net/wb6tpu
****

------------------------------

Date: Wed, 3 Jan 2001 17:01:10 -0600
From: Doug Hopperstad <[email protected]>
Subject: [SI-LIST] : RE: CDL file format

I asked my vender for IBIS or SPICE models for one of their devices
and was
sent a "CDL" formatted file. The email indicated that the file
contains all
the information about the sub-circuits and should be able to be
converted to
IBIS. Here is my question, what is a CDL file and has anyone converted
it to
either an IBIS or SPICE model? Any feedback would be greatly welcomed.

Here is some information I found on web regarding CDL files: A CDL
(network
Common data form Description Language) file is an ASCII description of
the
binary data in a netCDF file that is designed to be easily read by
humans.

Doug Hopperstad
Qlogic
[email protected]

**** To unsubscribe from si-list or si-list-digest: send e-mail to
[email protected]. In the BODY of message put: UNSUBSCRIBE
si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
si-list archives are accessible at http://www.qsl.net/wb6tpu
****

------------------------------

Date: Wed, 3 Jan 2001 18:00:03 -0500
From: "Ingraham, Andrew" <[email protected]>
Subject: RE: [SI-LIST] : Differential Impedance Effects on Diff.
Return Cu rrent

> First, is this statement valid?
> - In a differential pair, one 'leg' of a signal's return current
path is
> through the complementary 'leg' of a differential pair and not
through the
> ground or power planes (Assuming equal trace lengths, Zo=50
single-ended,
> Zo=100 diff. impedance - using ECL logic as an example).

The return current flows through any nearby conductors; at high
frequencies,
most of the return current is through the closest conductor(s).

If this happens to be a reference plane, that's where it goes. If it
is the
complementary signal line, that's where it goes.

You can construct a differential pair out of two separate traces that
are
far apart from one another, in which case each one has a return
current in
its neighboring plane(s) (which happens to roughly equal the signal
current
in the opposite trace). The two currents in the plane are opposite
and
would cancel at low frequencies, but at high frequencies the return
current
is confined to the space "under" each trace and does not cancel, so in
effect you have a circulating loop of return current in the planes, of
magnitude equal to the signal current.

In the other extreme, you can also construct a differential pair out
of two
traces that are tightly coupled to one another (such as broadside
coupled)
and further away from reference planes, in which case each line's
"return"
current is the other line's signal current and there is no significant
current in the planes.

Differential pairs have both a differential (odd-mode) impedance and a
common-mode (even-mode) impedance. The two examples could have the
same
differential impedance but very different even-mode impedances.

If you route two 50 ohm traces over ground planes, you'd get slightly
less
than 100 ohms differential impedance. To get closer to 100 ohms, you
need
to keep them further apart, in which case most of the return current
remains
in the planes. To get more cancellation of the plane's return
current, you
need them closer together, which reduces the differential impedance
(you
would need to use >50 ohm traces to get 100 ohms differential).

**** To unsubscribe from si-list or si-list-digest: send e-mail to
[email protected]. In the BODY of message put: UNSUBSCRIBE
si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
si-list archives are accessible at http://www.qsl.net/wb6tpu
****

------------------------------

Date: Wed, 03 Jan 2001 23:03:28 +0000
From: Martyn Gaudion <[email protected]>
Subject: Re: [SI-LIST] : Differential Impedance Effects on Diff.
Return Current

Dear Franz,

Yes you are correct, if a differential pair are routed with very close
spacing the return current
is through the other "leg" of the pair.

Move the spacing further apart and if you are between planes you reach
a point
where there is no coupling between the pair - and all the return
current is
through the ground and power planes. Once you reach this spacing you
have lost
many of the benefits of the differential pair.

You can see where the coupling reduces to zero by using a field solver
like Polar CITS25, - put some numbers in for a differential pair then
move
the pair further apart until they are uncoupled..

There is further information in the newly revised booklet on
controlled
impedance
PCBs which you can download as a PDF from www.polarinstruments.com

Kind regards
Martyn Gaudion
Polar Instruments
Tel + 44 1481 253081
email [email protected]

At 14:20 03/01/01 -0800, you wrote:
>All,
>
>Please excuse my 'newbie-like' questions, my inclusion in this
mailing list
>is more for curiosity and personal advancement/understanding than as
a
>profession.
>
>First, is this statement valid?
>- In a differential pair, one 'leg' of a signal's return current path
is
>through the complementary 'leg' of a differential pair and not
through the
>ground or power planes (Assuming equal trace lengths, Zo=50
single-ended,
>Zo=100 diff. impedance - using ECL logic as an example).
>
>Now, assuming the above statement is true:
>If the differential impedance is NOT 100 Ohms (Differential traces
NOT
>routed differentially) how does this effect the return current path?
Does
>the return current begin to flow through the ground and power planes
rather
>than through the differential pair?
>
>Thanks SI gurus!
>
>Franz.
>
>**** To unsubscribe from si-list or si-list-digest: send e-mail to
>[email protected]. In the BODY of message put: UNSUBSCRIBE
>si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
>si-list archives are accessible at http://www.qsl.net/wb6tpu
>****

**** To unsubscribe from si-list or si-list-digest: send e-mail to
[email protected]. In the BODY of message put: UNSUBSCRIBE
si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
si-list archives are accessible at http://www.qsl.net/wb6tpu
****

------------------------------

Date: Wed, 03 Jan 2001 15:12:09 -0800
From: Doug Brooks <[email protected]>
Subject: Re: [SI-LIST] : Differential Impedance Effects on Diff.
Return Current

At 02:20 PM 1/3/01 -0800, you wrote:
>All,
>
>Please excuse my 'newbie-like' questions, my inclusion in this
mailing list
>is more for curiosity and personal advancement/understanding than as
a
>profession.
>
>First, is this statement valid?
>- In a differential pair, one 'leg' of a signal's return current path
is
>through the complementary 'leg' of a differential pair and not
through the
>ground or power planes (Assuming equal trace lengths, Zo=50
single-ended,
>Zo=100 diff. impedance - using ECL logic as an example).
>
>Now, assuming the above statement is true:
>If the differential impedance is NOT 100 Ohms (Differential traces
NOT
>routed differentially) how does this effect the return current path?
Does
>the return current begin to flow through the ground and power planes
rather
>than through the differential pair?

First, change the question slightly to: If the SINGLE ENDED TRACE
IMPEDANCES ARE NOT EXACTLY MATCHED (it doesn't particularly matter
what the
values are, it matters whether they are matched) then the currents
down
them won't be exactly equal (and opposite) and therefore the
difference in
current will flow through (through the device and to) another return
path
(presumably ground).
Doug Brooks

.
************************************************************
Doug Brooks' book "Electrical Engineering for the Non-Degreed
Engineer" is now available. See our web site for details.
.
Doug Brooks, President [email protected]
UltraCAD Design, Inc. http://www.ultracad.com

**** To unsubscribe from si-list or si-list-digest: send e-mail to
[email protected]. In the BODY of message put: UNSUBSCRIBE
si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
si-list archives are accessible at http://www.qsl.net/wb6tpu
****

------------------------------

Date: Wed, 3 Jan 2001 17:37:14 -0600
From: "Chan, Michael" <[email protected]>
Subject: RE: [SI-LIST] : Differential Impedance Effects on Diff.
Return Cu rrent

Andy:
     If you have a tightly coupled pairs but there is also a reference
plane
near by
and if in this case the separation between the pair is about the same
as the
distance
of the traces to the reference plane then will the image current for
each
trace still
shows up on the trace or they will show up mostly on the reference
plane?

Regards,
Michael Chan

- -----Original Message-----
From: Ingraham, Andrew
Sent: Wednesday, January 03, 2001 5:00 PM
To: 'Dill, Franz @ Celerity'
Cc: 'SI-Mailing List'
Subject: RE: [SI-LIST] : Differential Impedance Effects on Diff.
Return
Cu rrent

> First, is this statement valid?
> - In a differential pair, one 'leg' of a signal's return current
path is
> through the complementary 'leg' of a differential pair and not
through the
> ground or power planes (Assuming equal trace lengths, Zo=50
single-ended,
> Zo=100 diff. impedance - using ECL logic as an example).

The return current flows through any nearby conductors; at high
frequencies,
most of the return current is through the closest conductor(s).

If this happens to be a reference plane, that's where it goes. If it
is the
complementary signal line, that's where it goes.

You can construct a differential pair out of two separate traces that
are
far apart from one another, in which case each one has a return
current in
its neighboring plane(s) (which happens to roughly equal the signal
current
in the opposite trace). The two currents in the plane are opposite
and
would cancel at low frequencies, but at high frequencies the return
current
is confined to the space "under" each trace and does not cancel, so in
effect you have a circulating loop of return current in the planes, of
magnitude equal to the signal current.

In the other extreme, you can also construct a differential pair out
of two
traces that are tightly coupled to one another (such as broadside
coupled)
and further away from reference planes, in which case each line's
"return"
current is the other line's signal current and there is no significant
current in the planes.

Differential pairs have both a differential (odd-mode) impedance and a
common-mode (even-mode) impedance. The two examples could have the
same
differential impedance but very different even-mode impedances.

If you route two 50 ohm traces over ground planes, you'd get slightly
less
than 100 ohms differential impedance. To get closer to 100 ohms, you
need
to keep them further apart, in which case most of the return current
remains
in the planes. To get more cancellation of the plane's return
current, you
need them closer together, which reduces the differential impedance
(you
would need to use >50 ohm traces to get 100 ohms differential).

**** To unsubscribe from si-list or si-list-digest: send e-mail to
[email protected]. In the BODY of message put: UNSUBSCRIBE
si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
si-list archives are accessible at http://www.qsl.net/wb6tpu
****

**** To unsubscribe from si-list or si-list-digest: send e-mail to
[email protected]. In the BODY of message put: UNSUBSCRIBE
si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
si-list archives are accessible at http://www.qsl.net/wb6tpu
****

------------------------------

Date: Wed, 3 Jan 2001 16:04:32 -0800
From: "Dill, Franz @ Celerity" <[email protected]>
Subject: RE: [SI-LIST] : Differential Impedance Effects on Diff.
Return Cu rrent

Andy,

Are the following observations then true?

If I wanted to isolate a ECL clock signal's return current path to the
diff.
complementary rail I would INCREASE the distance of both traces from
the
signal's reference plane but RETAIN the 100 Ohm differential impedance
and
terminate accordingly - making the diff. complementary rail the least
inductive path.
(Thinking about this - it's quite intuitive)

In a real-world situation where spacing to ref. plane is 2 mils
(Buried Cap)
with diff. spacing of 11 mils a majority of the return current will
run
through the ref. plane with a small amount through the complementary
diff.
rail. (Assumes no other near-by traces)

Franz.

- -----Original Message-----
From: Ingraham, Andrew [mailto:[email protected]]
Sent: Wednesday, January 03, 2001 3:00 PM
To: 'Dill, Franz @ Celerity'
Cc: 'SI-Mailing List'
Subject: RE: [SI-LIST] : Differential Impedance Effects on Diff.
Return
Cu rrent

> First, is this statement valid?
> - In a differential pair, one 'leg' of a signal's return current
path is
> through the complementary 'leg' of a differential pair and not
through the
> ground or power planes (Assuming equal trace lengths, Zo=50
single-ended,
> Zo=100 diff. impedance - using ECL logic as an example).

The return current flows through any nearby conductors; at high
frequencies,
most of the return current is through the closest conductor(s).

If this happens to be a reference plane, that's where it goes. If it
is the
complementary signal line, that's where it goes.

You can construct a differential pair out of two separate traces that
are
far apart from one another, in which case each one has a return
current in
its neighboring plane(s) (which happens to roughly equal the signal
current
in the opposite trace). The two currents in the plane are opposite
and
would cancel at low frequencies, but at high frequencies the return
current
is confined to the space "under" each trace and does not cancel, so in
effect you have a circulating loop of return current in the planes, of
magnitude equal to the signal current.

In the other extreme, you can also construct a differential pair out
of two
traces that are tightly coupled to one another (such as broadside
coupled)
and further away from reference planes, in which case each line's
"return"
current is the other line's signal current and there is no significant
current in the planes.

Differential pairs have both a differential (odd-mode) impedance and a
common-mode (even-mode) impedance. The two examples could have the
same
differential impedance but very different even-mode impedances.

If you route two 50 ohm traces over ground planes, you'd get slightly
less
than 100 ohms differential impedance. To get closer to 100 ohms, you
need
to keep them further apart, in which case most of the return current
remains
in the planes. To get more cancellation of the plane's return
current, you
need them closer together, which reduces the differential impedance
(you
would need to use >50 ohm traces to get 100 ohms differential).

**** To unsubscribe from si-list or si-list-digest: send e-mail to
[email protected]. In the BODY of message put: UNSUBSCRIBE
si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
si-list archives are accessible at http://www.qsl.net/wb6tpu
****

**** To unsubscribe from si-list or si-list-digest: send e-mail to
[email protected]. In the BODY of message put: UNSUBSCRIBE
si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
si-list archives are accessible at http://www.qsl.net/wb6tpu
****

------------------------------

Date: Wed, 03 Jan 2001 19:18:08 -0500
From: Michael Nudelman <[email protected]>
Subject: Re: [SI-LIST] : Differential Impedance Effects on Diff.
Return Current

Franz:

This depends how you routed your pairs.

If you have un-coupled pair (two independent transmission lines) - the
ret.
currents will flow into the ref. plane(s).

If the pair is tightly coupled - almost all current will flow in the
opposite
conductor.

If it is loosely coupled, the current will be shared between the plane
and the
opposite conductor.

I simulated some of the broadside-coupled pairs I use; when I removed
planes,
the impedance changed about 15%, which means also, that the pair was
tightly
coupled and most current would flow in the opposite conductor versus
the ref.
plane.

So, if you have 2 100Ohms lines, the resulting impedance is <= 50Ohm.
The
tighter the coupling, the lower the impedance will be.

And - the fast changing current always tries the least inductance
path. Which
alleviates understanding of where and why it will flow. For example,
all other
things being equal, the current will go in the closest parallel
conductor
(plane or wire) to minimize the loop area (the inductance is
proportional to
the loop area).

Mike.

"Dill, Franz @ Celerity" wrote:

> All,
>
> Please excuse my 'newbie-like' questions, my inclusion in this
mailing list
> is more for curiosity and personal advancement/understanding than as
a
> profession.
>
> First, is this statement valid?
> - In a differential pair, one 'leg' of a signal's return current
path is
> through the complementary 'leg' of a differential pair and not
through the
> ground or power planes (Assuming equal trace lengths, Zo=50
single-ended,
> Zo=100 diff. impedance - using ECL logic as an example).
>
> Now, assuming the above statement is true:
> If the differential impedance is NOT 100 Ohms (Differential traces
NOT
> routed differentially) how does this effect the return current path?
Does
> the return current begin to flow through the ground and power planes
rather
> than through the differential pair?
>
> Thanks SI gurus!
>
> Franz.
>
> **** To unsubscribe from si-list or si-list-digest: send e-mail to
> [email protected]. In the BODY of message put: UNSUBSCRIBE
> si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
> si-list archives are accessible at http://www.qsl.net/wb6tpu
> ****

**** To unsubscribe from si-list or si-list-digest: send e-mail to
[email protected]. In the BODY of message put: UNSUBSCRIBE
si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
si-list archives are accessible at http://www.qsl.net/wb6tpu
****

------------------------------

Date: Wed, 3 Jan 2001 16:23:01 -0800
From: "Dunbar, Tony" <[email protected]>
Subject: RE: [SI-LIST] : Differential Impedance Effects on Diff.
Return Cu rrent

Franz,

In the ideal world, the first statement is valid. It is also usually
the
intended operation that the designer is aiming for and assuming when a
(tightly-coupled) diff-pair is implemented. Any signal will always
find its
way to a destination via the path of least impedance. This fact is
most
often related to when considering the signal's return path. If the
return
path through the power/ground planes offers a lower impedance path
than any
other path, including the complementary trace of a diff-pair, then
that
least-impedance path will be used at any points along the way or way
back.

I think a key thing to keep in mind is to consider this in the
electromagnetic realm rather than simply the AC since it is the
electromagnetic coupling from the primary signal conductor to the
neigboring
conductive elements within the electromagnetic field that dictates
what the
return path elements will be.

I was going to move on and separately address your follow-up question
but I
think I may have covered that already with the answer above (??).

Regards,
Tony

- -----Original Message-----
From: Dill, Franz @ Celerity [mailto:[email protected]]
Sent: Wednesday, January 03, 2001 4:20 PM
To: 'SI-Mailing List'
Subject: [SI-LIST] : Differential Impedance Effects on Diff. Return
Current

All,

Please excuse my 'newbie-like' questions, my inclusion in this mailing
list
is more for curiosity and personal advancement/understanding than as a
profession.

First, is this statement valid?
- - In a differential pair, one 'leg' of a signal's return current
path is
through the complementary 'leg' of a differential pair and not through
the
ground or power planes (Assuming equal trace lengths, Zo=50
single-ended,
Zo=100 diff. impedance - using ECL logic as an example).

Now, assuming the above statement is true:
If the differential impedance is NOT 100 Ohms (Differential traces NOT
routed differentially) how does this effect the return current path?
Does
the return current begin to flow through the ground and power planes
rather
than through the differential pair?

Thanks SI gurus!

Franz.

**** To unsubscribe from si-list or si-list-digest: send e-mail to
[email protected]. In the BODY of message put: UNSUBSCRIBE
si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
si-list archives are accessible at http://www.qsl.net/wb6tpu
****

**** To unsubscribe from si-list or si-list-digest: send e-mail to
[email protected]. In the BODY of message put: UNSUBSCRIBE
si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
si-list archives are accessible at http://www.qsl.net/wb6tpu
****

------------------------------

Date: Wed, 3 Jan 2001 16:25:18 -0800 (PST)
From: Alan Zhu <[email protected]>
Subject: Re: [SI-LIST] : RE: CDL file format

Doug,

CDL stands for Circuit Description Language. It's meant to be used for
LVS (layout versus schematic) checking. It IS in Spice format. You can
use it directly in your Spice simulations. ( You may have to modify
some
of the parameters as some of them, for example the channel length, are
so
large that they are out of valid parameter range of the device model).

Alan Zhu
LightSand Communications

>From: Doug Hopperstad <[email protected]>
>To: [email protected]
>Subject: [SI-LIST] : RE: CDL file format
>Date: Wed, 3 Jan 2001 17:01:10 -0600
>MIME-Version: 1.0
>
>I asked my vender for IBIS or SPICE models for one of their devices
and was
>sent a "CDL" formatted file. The email indicated that the file
contains all
>the information about the sub-circuits and should be able to be
converted to
>IBIS. Here is my question, what is a CDL file and has anyone
converted it to
>either an IBIS or SPICE model? Any feedback would be greatly
welcomed.
>
>Here is some information I found on web regarding CDL files: A CDL
(network
>Common data form Description Language) file is an ASCII description
of the
>binary data in a netCDF file that is designed to be easily read by
humans.
>
>Doug Hopperstad
>Qlogic
>[email protected]
>
>
>
>
>**** To unsubscribe from si-list or si-list-digest: send e-mail to
>[email protected]. In the BODY of message put: UNSUBSCRIBE
>si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
>si-list archives are accessible at http://www.qsl.net/wb6tpu
>****
>

**** To unsubscribe from si-list or si-list-digest: send e-mail to
[email protected]. In the BODY of message put: UNSUBSCRIBE
si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
si-list archives are accessible at http://www.qsl.net/wb6tpu
****

------------------------------

End of si-list-digest V1 #377
*****************************

**** To unsubscribe from si-list or si-list-digest: send e-mail to
[email protected]. In the BODY of message put: UNSUBSCRIBE
si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
si-list archives are accessible at http://www.qsl.net/wb6tpu
****

**** To unsubscribe from si-list or si-list-digest: send e-mail to
[email protected]. In the BODY of message put: UNSUBSCRIBE
si-list or UNSUBSCRIBE si-list-digest, for more help, put HELP.
si-list archives are accessible at http://www.qsl.net/wb6tpu
****


About this list Date view Thread view Subject view Author view

This archive was generated by hypermail 2b29 : Tue May 08 2001 - 14:30:34 PDT