Re: [SI-LIST] : Trace impedance

Jim Lyke ([email protected])
Fri, 21 Aug 1998 07:05:48 -0600

Yes, conformal transforms for non-zero thickness conductors have been done
(see for example Chang,W.H. "Analytical IC Metal-Line Capacitance
Formulas," IEEE-MTT MTT-24(9):608-611, Sept 76). When you look at them,
you'll see why most people don't use them. If anyone's interested, I can
post a Mathematica script that implements some of these expressions.

As for measurement, you would absolutely suspect the measurements. I see a
spread in the predictions of the various formula. In some cases, my own
measurements fell in within the spread. I used a "driving-point" technique
for making the measurements, and my biggest uncertainties were the physical
dimensions, which I could not myself verify in my own work due to the
destructive methods I would have needed to employ. Keep in mind that there
is basically nothing in analytic solutions for the true inhomogeneous
dielectric smorgasbord that you usually encounter in life. As someone said
before, you can sometimes use fudge factors (e.g., an "effective"
dielectric constant to compensate for some effects, or "delta W", "delta H"
adjustments). Basically, you really can't casually approach this impedance
issue unless the application can accept casual answers.
jim

At 06:25 PM 8/20/98 -0700, Dima Smolyansky wrote:
>I believe there are some conformal mapping equations that actually take
>conductor thickness into account.
>
>Has anybody considered that the discrepancy between the calculation and
>measurement can actuall be due to poor measurements? I am not necessarily
>talk about impedance measurements (although, that too), but measurements
>and variations of board thickness and trace width, thickness etc.?
>
>Thanks,
>
>-Dima Smolyansky
>
>
>----------
>> From: Jim Lyke <[email protected]>
>> To: fabrizio=zanella%eng%[email protected]
>> Cc: casperdd <[email protected]>;
>[email protected]; [email protected]
><[email protected]>
>> Subject: Re: [SI-LIST] : Trace impedance
>> Date: Thursday, August 20, 1998 12:18 PM
>>
>> Most of these formulae were based on conformal transforms, often based on
>> zero thickness conductors. In a lot of modern work, those assumptions
>are
>> not very good, particularly when integrated circuits or multichip modules
>> are involved. I compared a lot of these expressions (Wheeler,
>Hammerstad,
>> Jensen, Chang, etc) against real-life measurements on 10u and 25u t-line
>> structures of different types in my Master thesis work and got results
>all
>> over the map.
>>
>> jim
>>
>> At 12:25 PM 8/20/98 -0400, fabrizio zanella wrote:
>> >Gentlemen, formulae for calculating impedances of stripline/microstrip
>> >traces are only somewhat accurate for certain geometries. The formulae
>> >were developed many years ago, when line widths were .010"+, dielectric
>> >thicknesses differed from today's designs, and most designs used 50 ohm
>> >transmission lines. For high impedance tlines, (80+ ohms), the accuracy
>of
>> >the formulae gets terrible.
>> >If you want to match your TDR measured impedances to calculations, you
>need
>> >to use a field solver tool. There are many good tools available that
>use
>> >boundary element or finite element analysis. Try Apsim's 'RLGC' tool,
>or
>> >Ansoft's Maxwell EZ2D.
>> >
>> >Regards,
>> >Fabrizio Zanella
>> >EMC Corporation
>> >[email protected]
>>
>> Jim Lyke
>> Air Force Research Lab
>> AFRL/VSSE
>> 3550 Aberdeen Ave SE
>> Bldg 887, Room 3
>> Kirtland AFB NM 87117-5776
>> 505-846-5812/5815
>>
>>
>> **** To unsubscribe from si-list: send e-mail to
>[email protected]. In the BODY of message put: UNSUBSCRIBE
>si-list, for more help, put HELP. si-list archives are accessible at
>http://www.qsl.net/wb6tpu/si-list ****
>
>
>
Jim Lyke
Air Force Research Lab
AFRL/VSSE
3550 Aberdeen Ave SE
Bldg 887, Room 3
Kirtland AFB NM 87117-5776
505-846-5812/5815

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