[SI-LIST] Re: Power plane noise measurement
- From: "istvan novak" <istvan.novak@xxxxxxxxxxxxxxxx>
- To: "Vishram Pandit" <vishrampandit@xxxxxxxxxxx>,<zhang_kun@xxxxxxxxxx>, <si-list@xxxxxxxxxxxxx>
- Date: Thu, 22 May 2003 22:10:20 -0400
Dear Vishram,
Using the capacitor pads for connection is a convenience, not a necessity. So
if you really want, you can put an SMA connector on the board. On the other
hand, after having done many measurements on power networks, experience tells
me that the SMA connector helps very little, it just takes up room on the
board. If your concern is that using a capacitor pad means that you have to
remove a capacitor, which may alter the results, use a pair of vias where
otherwise you would have the SMA connector. You can either solder cables to
the via pads or use small semirigid probes that fit into the holes so that you
can probe without any soldering.
If you have a large chip, making measurements in the pin field is not easy.
You can try using miniature probes, such as GGB Industries PicoProbes, or you
can put a pair of vias as close to the pin field as you can.
The value of the measured impedance depends on your design and on the
frequency. Low power circuits seldom require power-distribution impedance
below a few ohms. Low voltage and high-power systems, on the other hand, often
require a few milliohms or less at low and medium frequencies. At high
frequencies, self impedance will be dominated by inductance. If you use thin
laminates, the inductance may be as low as a few times ten pH.
Your 6-ohm impedance at 2GHz frequency translates into about 0.5 nH inductance.
Whether this is OK or not, depends on what transient noise current will enter
your power distribution network and how much noise voltage you can tolerate.
Regarding the highest frequency of interest, the rules may be different for SI
and EMI. For EMI purposes, you usually have to be concerned up to the first
four-to-six harmonics of the clock.
Best regards,
Istvan Novak
SUN Microsystems
----- Original Message -----
From: Vishram Pandit
To: istvan.novak@xxxxxxxxxxxxxxxx ; zhang_kun@xxxxxxxxxx ;
si-list@xxxxxxxxxxxxx
Sent: Wednesday, May 21, 2003 2:01 PM
Subject: Re: Power plane noise measurement
Dear Dr. Novak and Zangkun,
1] I can put an SMA connector (on a test board) at a less dense area of the
board. Typically there is a ground fill on the upper layer of the board, I can
connect the ground of the SMA connector to this ground fill. The center pin of
the SMA will go directly to the power plane through a via. Will this be okay?
Does it have to be at the capacitor pads?
2] Secondly, there may be different impedance response for power and ground
plane at different points on the board. If I am interested in measuring the
power/ground impedance at BGA pins, I may not be able to put an SMA connector
there. In this case, how to measure the impedance?
3] What are the typical values of the measured impedances? In my simulations,
I have got the following results:
* Best case board: from low frequency to 2GHz, power and ground impedance is
linearly rising up to 6 Ohms
* Worst case board: As high as 100 Ohms resonating at 800 MHz or beyond.
In simulations, I try to reduce it. What should be my criteria? Let us take
an example. The highest frequency on the board is 480MHz. I will try to reduce
the impedance upto 480MHz as far as possible. What should be my target
impedance at 480MHz? How far beyond 480MHz I can tolerate the resonances?
Any leads will be appreciated.
Regards,
Vishram Pandit
Senior Member Technical Staff
SI/EMC Engineering
Hughes Network Systems
Phone: (301)428-5725
>From: "istvan novak"
>To: ,,
>Subject: Re: [SI-LIST] Re: Power plane noise measurement
>Date: Wed, 21 May 2003 08:38:50 -0400
>
>Dear Zangkun and Vishram,
>
>The proper connection depends on what we want to measure.
>
>1., Connecting to pads of different caps, which are 'far' from
>each other in terms of wavelength of the highest frequency of
>interest, what we get is transfer impedance between the two
>connection points. The transfer impedance is more 'low-pass'
>in nature due to series lossesand possibly due to caps and
>other components on the board.
>2., Connecting to pads of the same capacitor, or to power/ground
>vias from the same side will give the self impedance of that point,
>which in this case is the self impedance of the power planes PLUS
>the via impedance in series. The series via impedance may mask
>out the details of the power/ground plane network.
>3., Connecting to the same via pair from the opposite sides will
>give the self impedance of the power-ground plane pairs alone by
>virtually eliminating the impedance of series via connections on
>either side. We can come close to this by connecting to pads of
>different capacitors, which are 'close'. If the highest frequency of
>interest is 1GHz, on an FR4 PCB 'close' could be 1cm or closer.
>
>Connecting to vias or pads of different 'close' capacitors on the
>same side of the PCB carries the additional risk of having so
>much coupling between the two connecting loops to the pads
>or vias that it again may mask out the impedance we want to
>measure. Connecting from opposite sides is a little bit trickier
>to make, but it gives the true self impedance of the structure
>without the added via impedance.
>
>Regards,
>
>Istvan Novak
>SUN Microsystems
>
>
>
>----- Original Message -----
>From: "Zhangkun"
>To: ; ;
>; ;
>Sent: Tuesday, May 20, 2003 11:20 PM
>Subject: [SI-LIST] Re: Power plane noise measurement
>
>
> >
> > Dear Vishram
> >
> > The two ports of VNA should connect to pads of different cap. Otherwise,
>the inductance of via and leading will overshadow the impedance of PDS. This
>is my experience:)
> >
> > Best Regards
> >
> > Zhangkun
> > 2003.05.21
> > ----- Original Message -----
> > From: "istvan novak"
> > To: ; ;
>;
> > Sent: Wednesday, May 21, 2003 9:02 AM
> > Subject: [SI-LIST] Re: Power plane noise measurement
> >
> >
> > >
> > > Vishram,
> > >
> > > Converting S11 values into impedance does work as long as you
> > > dont want to measure impedances below about 0.1 ohms. For
> > > low impedance values you still can use a network analyzer, but
> > > you have to connect both ports. You can search the archives of
> > > the si-list for past threads on the topic. Also, there was a half-day
> > > TecForum at DesignCon 2003 in Santa Clara earlier this year,
> > > where five companies presented power-distribution measurement
> > > techniques. Parts of it are available online, and an updated version
> > > of the TecForum will be presented late June in Boston, at
> > > DesignCon East 2003.
> > >
> > > regards,
> > >
> > > Istvan Novak
> > > SUN Microsystems
> > >
> > >
> > > ----- Original Message -----
> > > From: "Vishram Pandit"
> > > To: ; ;
> > > Sent: Tuesday, May 20, 2003 3:54 PM
> > > Subject: [SI-LIST] Re: Power plane noise measurement
> > >
> > >
> > > >
> > > >
> > > > Can anyone of you please elaborate about the measurement techniques?
>If I
> > > > want to measure
> > > >
> > > > 1] Time domain noise between power and ground,
> > > >
> > > > 2] Frequency domain Impedance between power and ground,
> > > >
> > > > what are the best ways of doing that?
> > > >
> > > > If I have a special SMA connector at the point of interest (between
>power
> > > > andground), I can use a network analyzer, to measure S11. Will this
be
>of
> > > > anyuse? Also, if I want to measure the impedance at different points
> > > > including the BGA pads, where it is very difficult to have special
> > > > connector,what is the best approach?
> > > >
> > > > Thanks for your reply,
> > > >
> > > > Vishram PanditSenior Member Technical Staff SI/EMC Engineering Hughes
> > > > NetworkSystems (301)428-5725
> > > >
> > > >
> > > >
> > > > >Virendra - To get a clean measurement of your power distribution
> >system,
> > > > you will want to remove the decoupling capacitor from it's pads
> >before
> > > > probing them. >>If the capacitor is left in place, it will "color"
the
> > > > measurement. The >decoupling capacitor mounted on the pads has a
>resonant
> > > > frequency. The >inductance for this resonant circuit is in the vias
>and
> > > > mounting pads >and in the height of the capacitor if you attach
probes
>to
> > > > thetop of >it. You will be measuring across the inductance which is
>highly
> > > > >involved in the resonant circuit and will measure mostly noise at
the
> > > > >"selfresonant frequency". >>Usually, there are many small capacitors
>on a
> > > > product. Pick one out of >a common value. For example, if there are
10
>x
> > > .01
> > > > uF caps, remove one >out of the 10. It will not have a large effect
on
>the
> > > > PDS that you are >measuring. >>regards, >Larry Smith >Sun
Microsystems
> > > > >>>Delivered-To: si-list@xxxxxxxxxxxxx >>Date: Tue, 20 May 2003
>10:28:36
> > > > -0700 (PDT) >>From: Virendra >>Subject: [SI-LIST] Power plane noise
> > > > measurement >>To: si-list@xxxxxxxxxxxxx >>MIME-Version: 1.0
> > > > >>Content-Transfer-Encoding: 8bit >>X-archive-position: 6885
> > > > >>X-ecartis-version: Ecartis v1.0.0 >>X-original-sender:
> > > adsurevv@xxxxxxxxx
> > > > >>X-list: si-list >>>>>>Hi all, >>>>i am trying to measure noise on a
>PCB
> > > > directly across a decoupling capacitor. >The PCB has buried/ blind
>vias
> > > and
> > > > has no specific power noise test points. My >question is, will there
>be a
> > > > difference in the measured noise if i measure noise >across the
>capacitor
> > > > andif i remove the capacitor and probe at the pads? I >cannot remove
>the
> > > > capacitor to do a what-if as i have only one board. >>>>thanks in
>advance,
> > > > >>>>Virendra Adsure >>>>>>>>--------------------------------- >>Do
you
> > > > Yahoo!? >>The New Yahoo! Search - Faster. Easier. Bingo.
> > > >
>>>>------------------------------------------------------------------
> > > >>To
> > > > unsubscribe from si-list: >>si-list-request@xxxxxxxxxxxxx with
> > > 'unsubscribe'
> > > > in the Subject field >>>>or to administer your membership from a web
>page,
> > > > goto: >>//www.freelists.org/webpage/si-list >>>>For help:
> > > > >>si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
> >>>>List
> > > > archives are viewable at: >>//www.freelists.org/archives/si-list
> >>or
> > > > atour remote archives:
> >>http://groups.yahoo.com/group/si-list/messages
> > > > >>Old(prior to June 6, 2001) list archives are viewable at:
> > > > >>http://www.qsl.net/wb6tpu
> > > >
> >>>>>>------------------------------------------------------------------
> > > >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: >//www.freelists.org/webpage/si-list >>For help:
> > > > >si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
>>List
> > > > archives are viewable at: >//www.freelists.org/archives/si-list
> >or
> > > at
> > > > our remote archives: >http://groups.yahoo.com/group/si-list/messages
> >Old
> > > > (prior to June 6, 2001) list archives are viewable at:
> > > > >http://www.qsl.net/wb6tpu >>
> > > >
> > >
> >
--------------------------------------------------------------------------
> > > --
> > > > The new MSN 8:[1] smart spam protection and 2 months FREE*
> > > >
> > > > --- Links ---
> > > > 1 http://g.msn.com/8HMJENUS/2737??PS=
> > > > ------------------------------------------------------------------
> > > > 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:
> > > > //www.freelists.org/webpage/si-list
> > > >
> > > > For help:
> > > > si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
> > > >
> > > > List archives are viewable at:
> > > > //www.freelists.org/archives/si-list
> > > > or at our remote archives:
> > > > http://groups.yahoo.com/group/si-list/messages
> > > > Old (prior to June 6, 2001) list archives are viewable at:
> > > > http://www.qsl.net/wb6tpu
> > > >
> > > >
> > >
> > > ------------------------------------------------------------------
> > > 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:
> > > //www.freelists.org/webpage/si-list
> > >
> > > For help:
> > > si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
> > >
> > > List archives are viewable at:
> > > //www.freelists.org/archives/si-list
> > > or at our remote archives:
> > > http://groups.yahoo.com/group/si-list/messages
> > > Old (prior to June 6, 2001) list archives are viewable at:
> > > http://www.qsl.net/wb6tpu
> > >
> > >
> > >
> > ------------------------------------------------------------------
> > 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:
> > //www.freelists.org/webpage/si-list
> >
> > For help:
> > si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
> >
> > List archives are viewable at:
> > //www.freelists.org/archives/si-list
> > or at our remote archives:
> > http://groups.yahoo.com/group/si-list/messages
> > Old (prior to June 6, 2001) list archives are viewable at:
> > http://www.qsl.net/wb6tpu
> >
> >
>
------------------------------------------------------------------------------
MSN 8 helps ELIMINATE E-MAIL VIRUSES. Get 2 months FREE*.
------------------------------------------------------------------
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:
//www.freelists.org/webpage/si-list
For help:
si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
List archives are viewable at:
//www.freelists.org/archives/si-list
or at our remote archives:
http://groups.yahoo.com/group/si-list/messages
Old (prior to June 6, 2001) list archives are viewable at:
http://www.qsl.net/wb6tpu
Other related posts:
