From: Larry Miller ([email protected])
Date: Fri Jun 09 2000 - 13:49:30 PDT
I do not see the connection (pardon the pun) between your references below
and measuring PCB impedance.
Our network analyzers have a 50-ohm impedance on the inputs and outputs. In
order to measure return loss (S11) on another impedance-level system you
need some kind of matching network. For example, we often have to measure
S11 for Ethernet Unshielded Twisted Pair (UTP, Category 5 cable) systems. To
do this we have to match the 100 ohm impedance of the system under test to
the 50 ohm impedance of the analyzer. We usually use precision baluns to do
this, since the 100 ohm system is also balanced to ground and the 50-ohm
tester is coax-- unbalanced.
Dr Johnson's point was that the PCB impedance is so low relative to any
feasible S11 test instrument that you have an impossible situation coupling
the instrument to the board. The impedance mismatch is so great that the
directional couplers used for an S11 measurement will not work correctly.
Think about how you calibrate a network analyzer: you apply 1) an
open-circuit, 2) a short circuit and finally 3) a (presumably) matched load.
In the case of a PCB, at the frequencies of interest there is virtually no
difference between the short circuit (2) and the matched load (3) (compared
to 50 ohms), so your measurement will have very poor resolution.
However, you can easily measure the transfer V-I characteristic. (I forget
whether this is actually S12 or not). Since the generator is operating into
a virtual short circuit the input current is just the generator voltage
divided by 50 ohms, and you measure the resulting voltage. Whatever you call
this (transfer impedance or whatever), it is easily measured over frequency
by a decent network analyzer. (I say "decent" because the limiting factor is
the receiver signal-to-noise ratio.)
I too do not know exactly why a time-domain instrument was mentioned, other
than it would let you use an oscilloscope as a receiver.
> -----Original Message-----
> From: Barry Ma [SMTP:[email protected]]
> Sent: Friday, June 09, 2000 8:56 AM
> To: Miller, Larry [SC7:322:EXCH]
> Cc: [email protected]
> Subject: RE: [SI-LIST] : Upper limit of interplane capacitance
> Yes, you are right. Many people investigate their PCB by using Vector
> Network Analyzer (VNA) with two cables. In the article you forwarded,
> Howard Johnson suggested using VNA in Time Domain to take place of TDR
> (Time Domain Reflector). Todd Hubing et al used VNA, too, in their paper
> "An Experimental Investigation of 4-layer PCB Decoupling", EMC Symposium
> 1995, pp. 308-312.
> In our case measuring the frequency response of pwr/gnd planes of PCB, we
> would like also to use VNA. What I suggested is to measure S11 with VNA by
> soldering a probe cable directly to two very closed Vcc and Gnd bonding
> pads on the PCB. (see the Figure in EDN magazine:
> http://www.ednmag.com/ednmag/reg/2000/03302000/pdfs/07john1.pdf ).
> Having connected the cable to Port 1 of VNA, the planecap becomes an
> unknown load being tested. The S11, thus the impedance of the interplane
> capacitance of pwr/gnd planes, would be shown in a Smith Chart in the VNA
> screen over a wide frequency range.
> BTW, a hand-held battery-powered (VNA + Spectrum Analyzer) has been on the
> market. Want to know the detail? Send me a short note off-line to me at
> "[email protected]".
> From the frequency response of S11 in the Smith chart, we are able to tell
> what is the upper limit for the operating frequency of interplane
> capacitance. IMHO, we don't need to measure S21 or S12, as Dr. Johnson and
> Dr. Hubing did in their articles. They investigated transmission
> properties along a PCB using two cables, one sending signal from Port 1 of
> VNA, and the other receiving signal for Port 2 of VNA. For our simpler
> task, however, measuring S11 with only one cable is enough.
> In addition to the planecap, we may use the hand-held (VNA+SA) to
> investigate many other important features of power distribution system in
> the PCB in the one-cable configuration, as shown in the Figure:
> - Vcc noise. (See
> - Can we reduce the decoupling capacitor number? People usually put too
> many. ...
> - Appropreate locations of decap. Closer is better? Or can be shared by
> other ICs.
> - What is the boundary of planecap and decaps.
> How to use VNA in Time Domain, as suggested by Howard Johnson, is also an
> interesting topic. I'd like to invite a discussion either on-line or
> Barry Ma [email protected]
> Anritsu www.anritsu.com
> Morgan Hill, CA 95037
> Tel. 408-778-2000 x 4465
> Best Regards,
> Barry Ma <[email protected]>
> ANRITSU www.anritsu.com
> Morgan Hill, CA 95037
> Tel. 408-778-2000 x 4465
> Why pay when you don't have to? Get AltaVista Free Internet Access now!
> **** 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
**** 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
This archive was generated by hypermail 2b29 : Wed Nov 22 2000 - 10:50:34 PST