Bob,
Just a couple of comments/corrections on your message.
Harry Parkinson
Chairman, IPC Controlled Impedance Task Group
Digital Equipment Corp
Tel 603-884-6760
FAX 603-884-1036
E-mail [log in to unmask]
From: US2RMC::"[log in to unmask]" "MAIL-11 Daemon" 28-FEB-1996 22:03:26.21
To: [log in to unmask], [log in to unmask]
CC:
Subj: Re: Re[2]: TDR Measured Dielectric Constant
I am not sure I have the terminology straight, and I would appreciate a
correction if I am wrong, but the following is my understanding:
Dual stripline refers to the case where two signal layers are placed between a
pair of ground planes. The signal line are generally orthogonal to avoid
cross talk. Impedances for this configuration can be measured with a standard
TDR. The equation to predict impedance is similar to but differs slightly from
the case of a single signal line between layers.
>> All single ended transmission lines, either stripline or microstrip are
>>unbalanced. Dual striplines can be either unbalanced (single ended) or
>>balanced (double ended or differential)
Balance strip lines refers to the case where signal are routed as "balanced
pairs" with one line for the outgoing pulse and one for the return pulse. The
impedance for this configuration is approximately (but not exactly) twice the
impedance of a single ended line. The problem is that a standard TDR cannot
measure the impedance for this configuration. The only choice is to measure
the two lines separately and use a formula to calculate impedance or buy a
special purpose test set.
>> TDR instrumentationCAN measure balanced transmission lines without
>>calculation I know of least two measuring systems, Tektronix 11801B and
>>Polar Instruments system (Channel Islands, U.K.). To measure the balanced ,
>>you connect a sampling probe to each line simultaneously. The formula
>> you refer to is (1+p (rho)/1-p(rho))times 100 (for a 50 ohm cable sampling
>>system)
With respect to copper foil, I would expect little effect other than a
possible reduction in effective spacing for a rough surface. At 1GHz the
effective wavelength is approximately 6 inches. Generally features less than
a quarter wavelength (1.5 inch) have little effect on the signal. Since the
topology is very much smaller than this, the electric field will effectively
average out the peaks and valleys in the conductor surface.
Bob Holmes
Lucent Technology (Formerly AT&T)
[log in to unmask]
>> An IPC document, IPC 2141, "Controlled Impednace Circuit Boards and High
>> Speed Logic Design will be published in the next month or so. It treats the
>> balanced transmission line case including discussion of the differences
>> between unbalanced and balanced lines, shows configuraions, formulas, and
>> measurement tecniques/configurations.
Harry Parkinson
------------- Begin Original Message -------------
From: hadco.com!dhoover
Date: Tue Feb 27 10:59:00 EST 96
Subject: Re[2]: TDR Measured Dielectric Constant
To: [log in to unmask]
Content-Type: text
Content-Length: 2512
Referring to FR-4.
Does Dual Stripline give you values that are similar to stripline?
Or should the measured sample be symmetrically balanced stripline?
With the measurement being ~ at 1 Ghz, can the copper foil topography
of the signal effect the Er calculated values?
(i.e., standard EDC vs VLP foils, thick black oxide vs reduced thin
oxide? If so, is it substantial?)
(BTW-I know that the copper foil subject has a major impact with
microwave due to capacitance variance)
Groovy
______________________________ Reply Separator
_________________________________
Subject: Re: TDR Measured Dielectric Constant
Author: [log in to unmask]
Date: 2/27/96 9:15 AM
The TDR is an excellent way to measure dielectric constant. With a known
configuration (stripline) and a standard formula for impedance one can easily
deduce the dielectric constant. Or more directly, propogation velocity is
easily measured and converted to dielectric constant. These methods generally
give dielectric constant values in the range of 4.2 for standard FR-4. This
is lower than the literature values of 4.6 to 4.8, but the literature values
are measured using a single frequency at 1 MHz while the TDR measures over a
range of frequencies that are at or above 1 GHz. In fact, dielectric constant
values provided by the TDR agree with frequency domain measurements made at 1
GHz. It should also be pointed out that real circuits in which impedance is
an issue generally operate at very high frequency, and the TDR values are much
more relevant than values at 1 MHz.
Nick Paulter of NIST (301-975-2405) in Gaithersburg Md has made an extensive
study in the use of TDR to measure dielectric constants. He is working on a
method which can be used on un-patterned raw material. I am sure he would be
interested in discussing his work with interested parties.
Bob Holmes Phd
Lucent Technology (Formerly AT&T)
[log in to unmask]
------------- Begin Original Message -------------
Has anyone attempted to use a Time Domain Reflectometer (TDR) to
measure the "effective dielectric" constant of internal Stripline and
Dualstripline signals?
Are the calculated dielectric constants close to what the laminate
supplier states or are they higher for FR4.
Richard Nolan
RGNolan Consulting
607-754-0079
% ====== Internet headers and postmarks (see DECWRL::GATEWAY.DOC) ======
% Received: from mail11.digital.com by us2rmc.zko.dec.com (5.65/rmc-22feb94) id AA22551; Wed, 28 Feb 96 21:43:42 -050
% Received: from IPC.ORG by mail11.digital.com (5.65v3.2/1.0/WV) id AA18942; Wed, 28 Feb 1996 21:44:50 -050
% Received: from ipc.org by simon.ipc.org via SMTP (940816.SGI.8.6.9/940406.SGI) id SAA24232; Wed, 28 Feb 1996 18:17:06 -080
% Resent-Date: Wed, 28 Feb 1996 18:17:06 -0800
% Received: by ipc.org (Smail3.1.28.1 #2) id m0trxZW-0000DmC; Wed, 28 Feb 96 19:54 CS
% Old-Return-Path: <[log in to unmask]>
% Message-Id: <[log in to unmask]>
% Date: 28 Feb 96 08:37:55 -0500
% To: [log in to unmask], [log in to unmask]
% From: [log in to unmask] (R_R_HOLMES)
% Original-Date: Wed Feb 28 08:37:55 EST 1996
% Phone: 804-226-5114
% Subject: Re: Re[2]: TDR Measured Dielectric Constant
% Original-To: hadco.com!dhoover
% Original-To: [log in to unmask]
% In-Reply-To: your message of Tue Feb 27 10:59:00 EST 96
% Content-Type: Text
% Resent-Message-Id: <"LPjJz3.0.D5B.DTGDn"@ipc>
% Resent-From: [log in to unmask]
% X-Mailing-List: <[log in to unmask]> archive/latest/2934
% X-Loop: [log in to unmask]
% Precedence: list
% Resent-Sender: [log in to unmask]
|