[SI-LIST] : Differential TDR probing

About this list Date view Thread view Subject view Author view

From: Eric Bogatin ([email protected])
Date: Thu Oct 26 2000 - 14:17:26 PDT

Hi folks-

I wanted to pull together a couple of comments to recent postings about TDR
measurements on differential pairs. Here they are in no particular order:

1. You can certainly measure the differential impedance of a pair of lines
using a single ended TDR. Each of the transmission lines in the pair has a
signal trace and a return path trace (often a plane in a MLB). You connect
the center pin of the TDR probe to the signal trace of one line and the
return path pin of the TDR probe to the second signal trace. What you
measure with the single channel TDR is the impedance the signal sees as it
propagates down the two signal traces. This is exactly the same waveform as
the difference signal if the two lines were driven differentially. You are
injecting a pure differential signal, and the differential impedance is,
after all, the impedance the difference signal sees. The return path planes
for the two transmission lines just sit there and float. I have an example
of this in the DesignCon 2000 paper, posted on my web site. However, you are
leaving a lot of potentially useful information on the table if this is all
you do.

2. As Dima from TDA Systems has recently pointed out, you can use a dual
head TDR from Agilent or Tek, for example, to drive two independent
transmission lines with correlated signals and measure more than just the
differential impedance. Each channel is a complete, independent TDR, with
steps that can be carefully adjusted for relative skew. Both scopes allow
the simultaneous output of a differential signal (0 to 400 mV on one channel
and 0 to -400 mV on the other channel for the Agilent scope) or a common
mode signal (0 to 400 mV in both channels in the Agilent scope) When the
scope is set to output a differential signal, you are driving the pair of
lines in the odd mode and each channel of the TDR is measuring the odd mode
impedance of the line it is connected to. When you drive both channels with
a common signal, each TDR channel is measuring the even mode impedance of
each line. In either case, each channel is driving the line it is connected
to just as though it were a single ended line- i.e., the center pin of the
TDR of one channel is connected to the signal path of one line and the
return pin of that TDR channel is connected to the return path of its
transmission line. It just happens that the two transmission lines in the
pair may have some coupling between them.

From the measured odd and even mode impedance of each line, you can extract
each of the three unique characteristic impedance matrix elements: Z11, Z22,
Z12=Z21 for the pair. The differential impedance of the line is Zodd1 +
Zodd2 or Z11-Z21 + Z22-Z12. If they are symmetric, then the differential
impedance is 2 x Zodd or 2(Z11-Z21). The common mode impedance is the
impedance the common signal sees when you drive the lines with the common
signal. This is the impedance of both lines, in parallel, or Zcommon = 1/2
Zeven, if they are symmetric. So there is a lot more you get when you
connect the two TDR channels up to the two coupled transmission lines.

3. I recently spent time at a Tek scope seminar and was very impressed with
the new high end scopes and TDR they have. However, they made the repeated
point that the Agilent dual TDR scope does not output simultaneous signals
in the two channels. Even after I politely said I was pretty sure it does,
their technical expert absolutely insisted that it did not. I have heard
this claim repeated on the SI list. The last time I looked at the waveforms
coming out of the dual heads (this morning), and based on the measurements I
have taken on differential pairs, I have to publicly challenge this comment
from Tek. The Agilent dual TDR does output simultaneous steps, and you can
manually adjust the phase shift between them to either get them to align to
within a few psec, or purposely add a skew.

4. One of the very useful functions a dual TDR scope and dual high bandwidth
amplifier plug in can do for you is allow you to emulate a fast rise time
differential signal and look at the far end at how much common mode signal
is generated. In the real world, in addition to driver skew, any asymmetry
in the electrical length of the interconnect, such as physical length,
loading of one line by test leads, pads, adjacent metal, etc., will generate
a common mode signal. With the dual channel TDR to generate the differential
signal, and a high bandwidth amplifier to receive the two channels at the
far end, you can directly measure the common mode converted just by the
interconnect. At the same time, you can add a skew delay between the two TDR
channels and look at the impact on generated common mode signal. An example
of doing this is in the DesignCon 2000 paper.

5. It has been brought to my attention that in some of the papers I've
posted on my web site, I have the common mode impedance listed as Zcommon =
Zeven. This is incorrect. The impedance the common mode signal sees is the
parallel combination of the two even mode impedances of each line in the
pair. The Zcommon is 1/2 Zeven, if the pair is symmetric. I have been so
focused on the differential impedance parts of my papers, I did not pay
close enough attention to what I had written about the common mode
impedance, and let this error slip through. I appreciate the feedback from
John Lin in particular, for pointing this out to me.

as always, I welcome comments.


Eric Bogatin
Training for Signal Integrity and Interconnect Design
v: 913-393-1305
f: 913-393-1306
e: [email protected]
web: <http://www.bogatinenterprises.com/>
ftp: ftp://ftp.BogatinEnterprises.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

About this list Date view Thread view Subject view Author view

This archive was generated by hypermail 2b29 : Tue May 08 2001 - 14:29:52 PDT