FreeLists Mailing List Manager wrote: > > si-list Digest Wed, 24 Jul 2002 Volume: 02 Issue: 201 > > In This Issue: > [SI-LIST] Re: doubt about crosstalk. > [SI-LIST] Re: doubt about crosstalk. > [SI-LIST] Re: doubt about crosstalk. > [SI-LIST] Re: doubt about crosstalk. > [SI-LIST] dont understand > [SI-LIST] Re: doubt about crosstalk. > [SI-LIST] dont understand > > ---------------------------------------------------------------------- > > Subject: [SI-LIST] Re: doubt about crosstalk. > From: "Gregory R Edlund" <gedlund@xxxxxxxxxx> > Date: Wed, 24 Jul 2002 08:32:21 -0500 > > Yun, > > Good questions. Let me take a stab at the first question. You made the > statement, "the ground guard line...has no influence on the magnetic > field." Let's try a thought experiment. Consider two parallel microstrip > lines WITHOUT a guard. Each line has a current flowing in it, a return > current in the plane, and a corresponding magnetic flux loop formed by the > current and it's return. Call these two loops Loop1 and Loop2. When > current in Loop1 changes, it creates a change in magnetic flux through > Loop2, and the changing flux induces a voltage on the second signal line. > Now, imagine that you keep the two signal lines at the same spacing but > introduce a ground guard trace between them. Because this guard trace is > carrying some of the return current from signal 1 (and signal 2), it > changes the shape of Loop1 and INTERCEPTS flux lines that would have > crossed into Loop2 had the guard not been there, thereby lowering the > induced voltage on the second signal line. Since there is no plane above > the structure, some of the flux lines will still go over the top and wind > up in Loop 2. > > So my conclusion is that a coplanar wave guide formed using a guard trace > effects both the electric AND magnetic coupling. I won't comment on > whether you're better off dropping the guard trace and using the same > spacing; I would defer that answer to a good field solver. I have heard > some folks say that the effective use of a guard trace is dependent on > drilling vias in the guard trace at regular intervals. I've never > investigated this myself. > > Greg Edlund > Senior Engineer > Signal Integrity > IBM Engineering Solution Services > 3605 Hwy. 52 N, Dept. HDC > Rochester, MN 55901 > gedlund@xxxxxxxxxx > > Date: Tue, 23 Jul 2002 15:57:11 +0800 > From: yun <yunnan@xxxxxxxxxx> > Subject: [SI-LIST] doubt about crosstalk. > > Hi, > I encounted some doubt about crosstalk,hope someone can help me. > 1.about ground guard line. > To reduce long coplanar parellel's crosstalk,we can place a ground guard > line between them. > On my opinion,however,the ground guard line can only terminate the electric > filed,it has no influnce > on magnetic filed.As we know,the forward crosstalk nearly canceled because > the negative polarity > of electric filed couple and magnetic filed couple.So, the forward > crosstalk will be larger when use > the ground guard line than not use the ground guard line. > 2.back to back return current crosstalk. > The pcb layer configuration is sig1-gnd-sig2(just for describe > question),then,both sig1 and sig2 layer's > return currents will all on the same gnd layer.We know that the return > currents will creat crosstalk on signal > lines.So,will these return currents,created by each side of gnd > layer,crosstalk? > Thanks. > yun > > ------------------------------ > > Subject: [SI-LIST] Re: doubt about crosstalk. > Date: Wed, 24 Jul 2002 14:13:45 -0500 > From: "Ibrahim Khan" <ikhan@xxxxxxxxxxxxxxxxxxx> > > Greg, > I have a question here. If the return current is also flowing in the > guard trace then hav'nt we basically increased the loop size on both > traces. > > Thanks > Ibrahim Khan > > -----Original Message----- > From: Gregory R Edlund [mailto:gedlund@xxxxxxxxxx]=20 > Sent: Wednesday, July 24, 2002 8:32 AM > To: si-list@xxxxxxxxxxxxx > Subject: [SI-LIST] Re: doubt about crosstalk. > > Yun, > > Good questions. Let me take a stab at the first question. You made the > statement, "the ground guard line...has no influence on the magnetic > field." Let's try a thought experiment. Consider two parallel > microstrip lines WITHOUT a guard. Each line has a current flowing in > it, a return current in the plane, and a corresponding magnetic flux > loop formed by the current and it's return. Call these two loops Loop1 > and Loop2. When current in Loop1 changes, it creates a change in > magnetic flux through Loop2, and the changing flux induces a voltage on > the second signal line. Now, imagine that you keep the two signal lines > at the same spacing but introduce a ground guard trace between them. > Because this guard trace is carrying some of the return current from > signal 1 (and signal 2), it changes the shape of Loop1 and INTERCEPTS > flux lines that would have crossed into Loop2 had the guard not been > there, thereby lowering the induced voltage on the second signal line. > Since there is no plane above the structure, some of the flux lines will > still go over the top and wind up in Loop 2. > > So my conclusion is that a coplanar wave guide formed using a guard > trace effects both the electric AND magnetic coupling. I won't comment > on whether you're better off dropping the guard trace and using the same > spacing; I would defer that answer to a good field solver. I have heard > some folks say that the effective use of a guard trace is dependent on > drilling vias in the guard trace at regular intervals. I've never > investigated this myself. > > Greg Edlund > Senior Engineer > Signal Integrity > IBM Engineering Solution Services > 3605 Hwy. 52 N, Dept. HDC > Rochester, MN 55901 > gedlund@xxxxxxxxxx > > Date: Tue, 23 Jul 2002 15:57:11 +0800 > From: yun <yunnan@xxxxxxxxxx> > Subject: [SI-LIST] doubt about crosstalk. > > Hi, > I encounted some doubt about crosstalk,hope someone can help me. 1.about > ground guard line. > To reduce long coplanar parellel's crosstalk,we can place a ground > guard line between them. On my opinion,however,the ground guard line can > only terminate the electric filed,it has no influnce on magnetic > filed.As we know,the forward crosstalk nearly canceled because the > negative polarity of electric filed couple and magnetic filed couple.So, > the forward crosstalk will be larger when use the ground guard line than > not use the ground guard line. 2.back to back return current crosstalk. > The pcb layer configuration is sig1-gnd-sig2(just for describe > question),then,both sig1 and sig2 layer's return currents will all on > the same gnd layer.We know that the return currents will creat crosstalk > on signal lines.So,will these return currents,created by each side of > gnd layer,crosstalk? Thanks. yun > > ------------------------------------------------------------------ > 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: =20 > //www.freelists.org/archives/si-list > or at our remote archives: > http://groups.yahoo.com/group/si-list/messages=20 > Old (prior to June 6, 2001) list archives are viewable at: > http://www.qsl.net/wb6tpu > =20 > > ------------------------------ > > From: MikonCons@xxxxxxx > Date: Wed, 24 Jul 2002 16:20:31 EDT > Subject: [SI-LIST] Re: doubt about crosstalk. > > In a message dated 7/24/2002 6:37:22 AM Pacific Standard Time, > gedlund@xxxxxxxxxx writes: > > > So my conclusion is that a coplanar wave guide formed using a guard trace > > effects both the electric AND magnetic coupling. I won't comment on > > whether you're better off dropping the guard trace and using the same > > spacing; I would defer that answer to a good field solver. I have heard > > some folks say that the effective use of a guard trace is dependent on > > drilling vias in the guard trace at regular intervals. I've never > > investigated this myself. > > > > Guard traces are indeed very effective. I constructed, tested, and reported > on the effects of guard traces way back in '89 as part of a seminar. A series > of microstrips, guarded microstrips, striplines, and guarded striplines were > constructed at specified 50 Ohm trace levels. The center-to-center spacing of > the signal traces was maintained for all configurations and the trace widths > adjusted to achieve the targeted 50 Ohm level for all lines. The LEAST > beneficial effect for the guarded traces was observed (as expected) for > microstrip. HOWEVER, the crosstalk was typically reduced by 14 dB and the > radiated emissions by 6-8 dB. QED. > > I DID include vias spaced randomly (to avoid a ladder resonance) by 0.75 inch > to 1.25 inch apart. For maximum effect, I believe the addition of vias to > critical (not garden variety) signals will assure an improved isolation > between signals. The added vias force "quiet ground" points along the signal > path that can serve to better shield the critical signal traces. > > Mike > > Michael L. Conn > Owner/Principal Consultant > Mikon Consulting > > *** Serving Your Needs with Technical Excellence *** > > ------------------------------ > > Date: Wed, 24 Jul 2002 13:28:56 -0700 > From: Val Mandrusov <valman@xxxxxxxxx> > Subject: [SI-LIST] Re: doubt about crosstalk. > > Mike, > > Could you share the information of this study (1989) through SI people. > > Thanks > > At 16:20 24.07.2002 -0400, MikonCons@xxxxxxx wrote: > >In a message dated 7/24/2002 6:37:22 AM Pacific Standard Time, > >gedlund@xxxxxxxxxx writes: > > > > > > > So my conclusion is that a coplanar wave guide formed using a guard trace > > > effects both the electric AND magnetic coupling. I won't comment on > > > whether you're better off dropping the guard trace and using the same > > > spacing; I would defer that answer to a good field solver. I have heard > > > some folks say that the effective use of a guard trace is dependent on > > > drilling vias in the guard trace at regular intervals. I've never > > > investigated this myself. > > > > > > >Guard traces are indeed very effective. I constructed, tested, and reported > >on the effects of guard traces way back in '89 as part of a seminar. A series > >of microstrips, guarded microstrips, striplines, and guarded striplines were > >constructed at specified 50 Ohm trace levels. The center-to-center spacing of > >the signal traces was maintained for all configurations and the trace widths > >adjusted to achieve the targeted 50 Ohm level for all lines. The LEAST > >beneficial effect for the guarded traces was observed (as expected) for > >microstrip. HOWEVER, the crosstalk was typically reduced by 14 dB and the > >radiated emissions by 6-8 dB. QED. > > > >I DID include vias spaced randomly (to avoid a ladder resonance) by 0.75 inch > >to 1.25 inch apart. For maximum effect, I believe the addition of vias to > >critical (not garden variety) signals will assure an improved isolation > >between signals. The added vias force "quiet ground" points along the signal > >path that can serve to better shield the critical signal traces. > > > >Mike > > > >Michael L. Conn > >Owner/Principal Consultant > >Mikon Consulting > > > > *** Serving Your Needs with Technical Excellence *** > > > > > >------------------------------------------------------------------ > >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 > > > > Val Mandrusov - Sr. Compliance Enginneer > > CISCO Systems, Inc > 375 East Tasman Drive > San Jose, CA 95134 > M/S SJ-06/04 > 408-853-7415 Phone > 408-853-3872 FAX > > ------------------------------ > > Subject: [SI-LIST] Re: doubt about crosstalk. > Date: Wed, 24 Jul 2002 15:34:17 -0500 > From: "Ibrahim Khan" <ikhan@xxxxxxxxxxxxxxxxxxx> > > Mike, > Can you please share your study with this group. > > Thanks > Ibrahim Khan > > -----Original Message----- > From: MikonCons@xxxxxxx [mailto:MikonCons@xxxxxxx]=20 > Sent: Wednesday, July 24, 2002 3:21 PM > To: si-list@xxxxxxxxxxxxx > Subject: [SI-LIST] Re: doubt about crosstalk.=20 > > In a message dated 7/24/2002 6:37:22 AM Pacific Standard Time,=20 > gedlund@xxxxxxxxxx writes: > > > So my conclusion is that a coplanar wave guide formed using a guard=20 > > trace effects both the electric AND magnetic coupling. I won't=20 > > comment on whether you're better off dropping the guard trace and=20 > > using the same spacing; I would defer that answer to a good field=20 > > solver. I have heard some folks say that the effective use of a guard > > > trace is dependent on drilling vias in the guard trace at regular=20 > > intervals. I've never investigated this myself. > >=20 > > Guard traces are indeed very effective. I constructed, tested, and > reported=20 > on the effects of guard traces way back in '89 as part of a seminar. A > series=20 > of microstrips, guarded microstrips, striplines, and guarded striplines > were=20 > constructed at specified 50 Ohm trace levels. The center-to-center > spacing of=20 > the signal traces was maintained for all configurations and the trace > widths=20 > adjusted to achieve the targeted 50 Ohm level for all lines. The LEAST=20 > beneficial effect for the guarded traces was observed (as expected) for=20 > microstrip. HOWEVER, the crosstalk was typically reduced by 14 dB and > the=20 > radiated emissions by 6-8 dB. QED. > > I DID include vias spaced randomly (to avoid a ladder resonance) by 0.75 > inch=20 > to 1.25 inch apart. For maximum effect, I believe the addition of vias > to=20 > critical (not garden variety) signals will assure an improved isolation=20 > between signals. The added vias force "quiet ground" points along the > signal=20 > path that can serve to better shield the critical signal traces. > > Mike > > Michael L. Conn > Owner/Principal Consultant > Mikon Consulting > > *** Serving Your Needs with Technical Excellence *** > > ------------------------------------------------------------------ > 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: =20 > //www.freelists.org/archives/si-list > or at our remote archives: > http://groups.yahoo.com/group/si-list/messages=20 > Old (prior to June 6, 2001) list archives are viewable at: > http://www.qsl.net/wb6tpu > =20 > > ------------------------------ > > From: "Volk, Andrew M" <andrew.m.volk@xxxxxxxxx> > Subject: [SI-LIST] Re: doubt about crosstalk. > Date: Wed, 24 Jul 2002 14:05:36 -0700 > > Mike - > > I'm sure of the effectiveness of the guard strip, especially with > grounding vias, but vias require a lot of space. I was wondering > if you did a control with the same layout without the guard trace > and only used the spacing as the isolation? > > Regards, > Andrew Volk > > -----Original Message----- > From: MikonCons@xxxxxxx [mailto:MikonCons@xxxxxxx] > Sent: Wednesday, July 24, 2002 1:21 PM > To: si-list@xxxxxxxxxxxxx > Subject: [SI-LIST] Re: doubt about crosstalk. > > In a message dated 7/24/2002 6:37:22 AM Pacific Standard Time, > gedlund@xxxxxxxxxx writes: > > > So my conclusion is that a coplanar wave guide formed using a guard trace > > effects both the electric AND magnetic coupling. I won't comment on > > whether you're better off dropping the guard trace and using the same > > spacing; I would defer that answer to a good field solver. I have heard > > some folks say that the effective use of a guard trace is dependent on > > drilling vias in the guard trace at regular intervals. I've never > > investigated this myself. > > > > Guard traces are indeed very effective. I constructed, tested, and reported > on the effects of guard traces way back in '89 as part of a seminar. A > series > of microstrips, guarded microstrips, striplines, and guarded striplines were > > constructed at specified 50 Ohm trace levels. The center-to-center spacing > of > the signal traces was maintained for all configurations and the trace widths > > adjusted to achieve the targeted 50 Ohm level for all lines. The LEAST > beneficial effect for the guarded traces was observed (as expected) for > microstrip. HOWEVER, the crosstalk was typically reduced by 14 dB and the > radiated emissions by 6-8 dB. QED. > > I DID include vias spaced randomly (to avoid a ladder resonance) by 0.75 > inch > to 1.25 inch apart. For maximum effect, I believe the addition of vias to > critical (not garden variety) signals will assure an improved isolation > between signals. The added vias force "quiet ground" points along the signal > > path that can serve to better shield the critical signal traces. > > Mike > > Michael L. Conn > Owner/Principal Consultant > Mikon Consulting > > *** Serving Your Needs with Technical Excellence *** > > ------------------------------------------------------------------ > 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 > > > ------------------------------ > > From: MikonCons@xxxxxxx > Date: Wed, 24 Jul 2002 17:56:31 EDT > Subject: [SI-LIST] Re: doubt about crosstalk. > > Per multiple comments/requests: > > [Q] "I was wondering if you did a control with the same layout without the > guard trace > and only used the spacing as the isolation?" > > [Mikon] Yes. The reference point for performance was precisely what you > describe. As I originally stated, "A series of microstrips, guarded > microstrips, striplines, and guarded striplines were constructed at specified > 50 Ohm trace levels. The center-to-center spacing of the signal traces was > maintained for all configurations and the trace widths adjusted to achieve > the targeted 50 Ohm level for all lines." > > [Q] "There are two components to crosstalk, magnetic and capacitive. I can > see > where a guard trace can easily help with the capacitive crosstalk but the > effects on magnetic crosstalk would be much less. The magnetic field contains > energy in the flux lines. It is unclear to me whether the magnetic crosstalk > can > be lowered by placing a parasitic loop between the aggressor and the victim > traces. > > [Mikon] The guard traces (as well as the ground planes) carry high frequency > return currents (i.e., in the opposite direction) induced by the signals and > help reduce the magnetic loop and the field intensities created thereby. > > [Q] "Can you please share your study with this group." > > Actually, I thought I just did share my results, at least about the crosstalk > question. In the same tutorial, I also introduced the multiple-layer, > co-planar, via-connected, chassis ground trace techniques for reduction of > edge-related radiated emissions (and superior ESD protection), multiple other > radiation and crosstalk reduction techniques, and a dissertation on buried > capacitance versus discrete capacitors for power distribution system > decoupling. The full tutorial is over an hour long and has long since been > archived in my files. It was also in a different presentation format than the > now more common Microsoft PowerPoint files. > > You might try to access the files at Hewlett-Packard (now Agilent) as that > work is published multiple times in the 1990 to 1993 HP High-Speed Design > Seminar books. They also did a videotape of my associated presentation which > was marketed (via a third party) out of either San Antonio or Houston, Texas > (I don't precisely recall which). > > The most recent publication I know of is in both the U.S. and European HP > seminar books "1993 High Speed Digital Design Symposium." At that time, Jim > Kabel was the Test and Measurement U.S. Marketing Center Manager for HP. You > might try to locate more information through him. > > That's all I have time for now, so good luck. > > Mike > > Michael L. Conn > Owner/Principal Consultant > Mikon Consulting > > *** Serving Your Needs with Technical Excellence *** > > ------------------------------ > > Date: Thu, 25 Jul 2002 08:44:59 +0800 > From: qzheng <qzheng@xxxxxxxxxxxxxxxx> > Subject: [SI-LIST] dont understand > > Hello si-list, > > I dont understand the meaning of "ladder resonances" ?? who can > help > > thanks > > -- > Best regards, > qzheng mailto:qzheng@xxxxxxxxxxxxxxxx > > ------------------------------ > > From: "Ismail B - CTD, Chennai." <ismailb@xxxxxxxxxxxxxxx> > Subject: [SI-LIST] Re: doubt about crosstalk. > Date: Thu, 25 Jul 2002 10:42:26 +0530 > > >[Mikon] The guard traces (as well as the ground planes) carry high > frequency > >return currents (i.e., in the opposite direction) induced by the signals > and > >help reduce the magnetic loop and the field intensities created thereby. > > Mike, your's is a good explanation. In my opionion the ground planes carry > the same > high frequency return currents as the signal traces, but in an opposite > direction. > Hence the magnetic field created by the ground strips should oppose the > magnetic > field produced by the signal traces, thereby reducing the degree of > coupling. > > Regards, > Ismail > > ------------------------------ > > From: MikonCons@xxxxxxx > Date: Thu, 25 Jul 2002 01:19:09 EDT > Subject: [SI-LIST] dont understand > > Qzheng: > > Sorry about the confusion. I refer to an equally spaced row of vias as > representing a "ladder" with repetitive rungs. Such a structure will have a > common resonant frequency for each of the rung segments; therefore, I choose > to stagger the spaces between vias to detune possible reinforcement of a > given resonance. > > Mike > > Michael L. Conn > Owner/Principal Consultant > Mikon Consulting > > *** Serving Your Needs with Technical Excellence *** > > ------------------------------ > > End of si-list Digest V2 #201 > ***************************** ------------------------------------------------------------------ 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