[SI-LIST] Re: Differential TDR

  • From: Istvan Novak <istvan.novak@xxxxxxx>
  • To: zcheng@xxxxxxxxxxxxxxx
  • Date: Sun, 14 Dec 2008 18:04:57 -0500


As Yuriy also pointed out, you still will measure the differential 
response, even if there is no ground connection.  If we have a single 
'ground' connection for a balanced circuit, the 'ground' will have no 
fluctuation due to a differential excitation, so whether you connect it 
or not, it will not change the result.  Unbalanced circuits, however 
(for instance, if the two single-ended impedances are different, or if 
there is noticeable skew between the legs), will result in a residual 
signal on its 'ground' and therefore its stray coupling to cable and 
instrument ground will alter the result.  Bottom line: for well-balanced 
circuits and if you need only the differential response, the ground 
connection is not needed.  For all other cases you need the ground 


Istvan Novak
SUN Microsystems

> Dr. Novak,
> Thanks for your reply.
> If there is Gnd connection from differential probe to DUT, the 
> measured Zdiff will be still the same as that measured w/o gnd connection?
> My concern is: w/o gnd connection to DUT, the signal return path will 
> be each other, (i.e., p reference to n, and n reference to p); when 
> gnd is added, signal might partially refer to gnd, the return path 
> might be different from w/o gnd situation. Then will the Zdiff still 
> the same?
> ZG
> 2008/12/13 Istvan Novak <istvan.novak@xxxxxxx 
> <mailto:istvan.novak@xxxxxxx>>
>     ZG, Orin,
>     If the probe is really part of a TDR setup, I dont think
>     high-impedance probe is an option.
>     If the DUT is passive and well balanced, the differential probe
>     without ground connection
>     measured the differential response.  It can not, however, measure
>     the common-mode response
>     without a ground connection to the DUT.  So to answer the original
>     question, without a ground
>     connection to the DUT, only the differential impedance can be
>     measured.  And since without
>     ground connection the single-ended impedance can not be measured,
>     both Z_single and
>     Z_mutual will remain unknown.
>     Regards,
>     Istvan Novak
>     SUN Microsystems
>     olaney@xxxxxxxx <mailto:olaney@xxxxxxxx> wrote:
>         If you mean a hi-z diff probe, then the line needs to be
>         terminated with
>         matched resistors to ref ground.  If dual 50 ohm probes into
>         50 ohm
>         inputs, the instrument itself provides termination.  The
>         instrument
>         doesn't need to "know" what the single ended impedance is to
>         measure
>         Zdiff.  That's not to say that Zse or Zcm aren't also
>         available from the
>         typical setup.  The most important consideration for Zdiff
>         accuracy is
>         that measurement path delays be identical for both probes
>         (automatic for
>         a unitary diff probe), and that the probe grounds be tied to
>         each other
>         (not necessarily to other grounds) at the measurement end
>         (also automatic
>         for a unitary diff probe).  Also note that Zdiff can be well
>         defined even
>         if Zcm is not, hence the successful application of unshielded
>         twisted
>         pairs dangling in air and over heating ducts in the form of
>         CAT5 LAN
>         cabling.  The large swings in Zcm barely affect Zdiff.
>         Orin Laney
>         On Sat, 13 Dec 2008 12:22:54 -0800 "ZHENGGANG CHENG"
>         <zhenggang.cheng@xxxxxxxxx <mailto:zhenggang.cheng@xxxxxxxxx>>
>         writes:
>             Hi all,
>             When using differential probe to do the TDR for
>             differential pair, if the
>             probe only has P and N tip no Gnd tip, does the TDR gives
>             accurate values?
>             Here is the concern:
>             Zdiff =2*(Z_single-Z_mutual), if no Gnd tip is used, how
>             does the TDR know
>             what's the single-ended impedance, and consequently how
>             does TDR know the
>             Zdiff?
>             Thanks,
>             ZG

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:

For help:
si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field

List technical documents are available at:

List archives are viewable at:     
or at our remote archives:
Old (prior to June 6, 2001) list archives are viewable at:

Other related posts: