[SI-LIST] Re: The crosstalk of transmission line when it is completelymatched
- From: Matthew Herndon <mherndon@xxxxxxxxx>
- To: zhang_kun@xxxxxxxxxx
- Date: Mon, 9 Dec 2002 11:52:04 -0800
Zhangkun -
Your question is timely for me: on Friday, I submitted a paper for
January's DesignCon on the topic of Near-end crosstalk (NEXT), so I
have been thinking about these issues.
When I run the circuit in your first posting (Dec 5, 2002), I do indeed
get puzzling results - FEXT looks the same as NEXT, except delayed. To
check things out, I terminated each node of the transmission line
individually. When I did so, I got the results I (and the textbooks)
expect. See the PS below for exact details.
I am not familiar with the termination scheme you used on the far end.
Perhaps you (or someone on the list) can explain it to me, but I think
it is the problem.
On your question about FEXT: both inductive and capacitive coupling
cause crosstalk. These factors add for NEXT and subtract for FEXT. Good
references on this topic are: Hall, Hall and McCall "High-Speed Digital
System Design" and Young "Digital Signal Integrity". The relevant
formulas are (see PPS below for a definition of the symbols):
Vpeak(NEXT) = (Vin/4)*(Lm/L + Cm/C)
Vpeak(FEXT) = -(Vin*Tf/Tr)*(Lm/L - Cm/C)
In your case, Lm/L is 0.1868 and Cm/C is 0.1875, so FEXT is (nearly) 0.
I have been playing around with this phenomenon some. I only did a few
simulations, but the property holds for all stripline cases I tried. I
varied er, trace impedance, and even tried an asymmetrical stackup.
Does anyone know if this property holds for all stripline, and if so
why?
BTW, the property does not hold for microstrip.
-matt
Matt Herndon
ECAD Group
Apple Computer, Inc.
PS - Exact details and results of termination changes I made: I
terminated out1 with a 47.69 Ohms to ground, and did the same with
out2, and (taking Scott's suggestion) put a 47.69 Ohm resistor between
the source and in1. The results from this circuit are more what I (and
the textbooks) expect: Looking at in2 at the first NEXT pulse, which
starts at the same time as the agressor pulse (1000psec), we see it has
a rise time which is the same as the agressor (1000psec), then stays
flat ("saturated") until two flight times have elapsed from the start
of the agressor pulse (i.e., 1700 psec have elapsed), then returns to
zero after another 1000psec . The pulse has the same polarity as dV/dt
on the agressor. We see the same thing repeated on the downward
transition 1300psec later. If anyone wants my netlist, I can send it.
PPS - Symbols in the equations: Vin is the amplitude of the agressor
pulse (assumed PWL), Tf is the flight time, Tr is the rise time of the
aggressor pulse (assumed triangular), Lm is the inductive coupling of
the traces (6.40120e-08 in your rlgc file), L is the self-inductance
of the line (3.42622e-07), Cm is the capacitive coupling (2.82540e-11)
and C is the (total?) capacitance (I believe you can use 1.50678e-10
for this, there is some confusion about whether this number in HSPICE
is total capacitance or self-capacitance - perhaps someone can
clarify). Lm, L, Cm, and C are per unit length, of course.
On Sunday, December 8, 2002, at 06:53 PM, Zhangkun wrote:
>
> Hi McMorrow
>
> I do not agree with you. Please try the following netlist. The far-end
> crosstalk is zero.
>
> match_1.sp
> the simple diff transmissionline
> .options list node post
> .op
> .tran 0.5n 100n
> *driver
> vp in1 gnd pulse 0 5 1n 1n 1n 3n 8n
> *vn in2 gnd pulse 10 0 1n 1n 1n 3n 8n
> *transmission line
> Wcable in1 in2 gnd out1 out2 gnd RLGCfile=diffline.rlgc n=2 l=5000mil
> *receiver
> rin2 in2 gnd 47.69
> rout1 out1 gnd 47.69
> rout2 out2 gnd 47.69
> .end
>
> The difference is the terminal at the farend. There is no difference
> at near end.
>
> Best Regards
> Zhangkun
> 2002.12.09
>
> ----- Original Message -----
> From: Scott McMorrow <scott@xxxxxxxxxxxxx>
> To: <zhang_kun@xxxxxxxxxx>; <si-list@xxxxxxxxxxxxx>
> Sent: Friday, December 06, 2002 11:38 AM
> Subject: Re: [SI-LIST] The crosstalk of transmission line when it is
> completelymatched
>
>
>> Zhangkun,
>>
>> You are getting reflected reverse crosstalk off of the driver. You
>> will
>> need to match the driver impedance to the transmission line impedance
>> to
>> eliminate this reflection. Since the driver is a perfect source, your
>> reflection coefficient is -1.
>>
>> scott
>>
>> --
>> Scott McMorrow
>> Teraspeed Consulting Group LLC
>> 2926 SE Yamhill St.
>> Portland, OR 97214
>> (503) 239-5536
>> http://www.teraspeed.com
>>
>>
>>
>>
>>
>>
>> Zhangkun wrote:
>>
>>> Hi all:
>>>
>>> This morning, I do some crosstalk simulation. I find something
>>> interesting. The following is what I do.
>>>
>>> 1. Create diffline.rlgc from rlgc.sp
>>> 2. Calculate the Z_diff and Z_comm from diffline.rlgc
>>> 3. Match the diffline and simulate it by match_2.sp
>>>
>>> I find the near end crosstalk and the far end crosstalk are equal.
>>> This is not same to the general crosstalk theory. I want to know >>> why.
>>>
>>> rlgc.sp
>>> .OPTION PROBE POST
>>> VIMPULSE in1 gnd PULSE 0v 5v 5n 2n 2n 20n
>>> W1 in1 in2 gnd out1 out2 gnd FSmodel=cond2_sys N=2 l=1000mil
>>> Rin2 in2 gnd 49
>>> Rout1 out1 gnd 49
>>> Rout2 out2 gnd 49
>>> .MATERIAL diel DIELECTRIC ER=4.5
>>> .SHAPE rect RECTANGLE WIDTH=7mil, HEIGHT=2mil,
>>> .LAYERSTACK stack1,
>>> +LAYER=(PEC,20mil), LAYER=(diel,22mil),
>>> +LAYER=(PEC,20mil)
>>> .FSOPTIONS opt1 PRINTDATA=YES
>>> .MODEL cond2_sys W MODELTYPE=FieldSolver,
>>> +LAYERSTACK=stack1, FSOPTIONS=opt1
>>> +RLGCFILE=diffline.rlgc
>>> +CONDUCTOR=(SHAPE=rect, ORIGIN=(0,30mil)),
>>> +CONDUCTOR=(SHAPE=rect, ORIGIN=(14mil,30mil))
>>> .TRAN 0.5n 100n
>>> .PROBE v(out1)
>>> .END
>>>
>>> diffline.sp
>>> *SYSTEM_NAME : cond2_sys
>>> *
>>> * ------------------------------------ Z = 0.0015748
>>> * //// Top Ground Plane //////////////
>>> * ------------------------------------ Z = 0.0010668
>>> * diel H = 0.0005588
>>> * ------------------------------------ Z = 0.000508
>>> * //// Bottom Ground Plane ///////////
>>> * ------------------------------------ Z = 0
>>>
>>> * N (number of signal conductors)
>>> *********************************
>>> 2
>>>
>>> * Lo (H/m)
>>> *********************************
>>> 3.42622e-07
>>> 6.40120e-08 3.42622e-07
>>>
>>>
>>> * Co (F/m)
>>> *********************************
>>> 1.50678e-10
>>> -2.82540e-11 1.50678e-10
>>>
>>>
>>> * Ro (Ohms/m)
>>> *********************************
>>> 0.00000e+00
>>> 0.00000e+00 0.00000e+00
>>>
>>>
>>> * Go (S/m)
>>> *********************************
>>> 0.00000e+00
>>> -0.00000e+00 0.00000e+00
>>>
>>> match_2.sp
>>> the simple diff transmissionline
>>>
>>> .options list node post
>>> .op
>>> .tran 0.5n 100n
>>>
>>> *driver
>>> vp in1 gnd pulse 0 5 1n 1n 1n 3n 8n
>>> *vn in2 gnd pulse 10 0 1n 1n 1n 3n 8n
>>> rin2 in2 gnd 47.69
>>>
>>> *transmission line
>>> Wcable in1 in2 gnd out1 out2 gnd RLGCfile=diffline.rlgc n=2 l=5000mil
>>>
>>> *receiver
>>> rout1 out1 node1 39.4598
>>> rout2 out2 node1 39.4598
>>> rcon node1 gnd 9.0864
>>>
>>> .end
>>>
>>> Z_diff=78.9195 Z_comm=28.8163
>>> rou1=rout2=Z_diff/2
>>> rcon=(2*Z_comm-Z_diff/2)/2
>>>
>>> Best Regards
>>> Zhangkun
>>> 2002.12.06
>>>
>>>
>>> ------------------------------------------------------------------
>>> 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
>>>
>>>
>>>
>>>
>>>
>>
>>
>>
> ------------------------------------------------------------------
> 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
>
>
------------------------------------------------------------------
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
Other related posts:
