[SI-LIST] Re: Parallel resonance (anti-resonance) on power distribution system and Full impedance compensation concept

Steve, Rodrigo,

In the context how the Mezhiba, Friedman book treats it in Section 5.7, 
it is a derivative of the
Adaptive Voltage Positioning concept (by Redl et al) and Extended 
Adaptive Voltage Positioning
(by Waizman and Chung).  Figures 5.21 and 5.22 indicate the original 
intent of having so low
Q from each component bank that there is a flat impedance bottom before 
the next component
bank joins in on the frequency scale.

As Steve pointed out, a key requirement is to keep inductance low, but 
it is relative to SRF and
ESR: high ESR components work very well with moderate or high inductance 
in such designs.

Yes, the concept can be implemented with good results, as long as 
spatial variations are not
significant and/or suppressed.  This also means that this design 
methodology usually stops at
the package-silicon interface.  And yes, it in fact has been 
implemented; many thousands
boards and systems have been shipped, implementing different flavors of 
the concept.

When it comes to simulation tools: we need to remember that they are 
what they are: simulation
tools, analyzing the design we plug in.  In that regard any simulation 
tool can handle the concept.
I suspect the question was more about design features of the tools: 
several PDN tools today
offer  PDN optimization, and those optimization procedures in theory 
encompass  the full
impedance compensation variant as well.  The caveat is that the tool 
will give results based on
the component library it has built in, so you may need to add the very 
low Q parts and force
the tool to use those.

Regards,

Istvan Novak
Oracle-SUN




steve weir wrote:
> That term appears in Mezhiba, Friedman "Power Distribution Networks in 
> High Speed Integrated Circuits".  As used in that book, it refers to 
> tuned compensation networks in order to damp out PDN resonances.  In 
> effect, all multipole networks (FDTIM in Altera parlance) do this to 
> some degree although in more of a whack-a-mole fashion.  Sigrity's 
> Optimize PI(tm) product will calculate the capacitor network for you.  
> The Cadence Allegro PI(tm) and Mentor Hyperlynx PI(tm)tools will both 
> evaluate a capacitor network of your specification.
>
> In order to perform the kind of compensation that Mezhiba and Friedman 
> talk about, the tricky part is finding capacitors with the combination 
> of ESR and capacitance that you want.  As frequency goes up this becomes 
> tougher and tougher.  You are very unlikely to find the values you want 
> right off the shelf, forcing compromise.  If you pick a capacitor and 
> the mounted SRF lands in the wrong place due to tolerances, and/or the 
> capacitor Q is too high, then you will take one peak and turn it into 
> two or more peaks.  Since inductance directly contributes to Q, as in 
> all other things PDN, inductance is the primary enemy here. The lower 
> the mounted inductance of the cap compared to the ESR the easier it will 
> be to realize an effective compensation network.  Because they have very 
> low mounted inductance and are available over a wide value range, X2Y(r) 
> caps are particularly good for this kind of duty.  There is an example 
> on the X2Y(r) website where one capacitor compensated out a resonance at 
> about 600MHz very nicely contributing to dramatically improved SerDes 
> jitter:
>
> http://www.x2y.com/bypass/method/altera_bypass.pdf
>
> Steve.
>
>
> Rodrigo Rodriguez wrote:
>   
>> Hello,
>>
>>  
>>
>> >From what I read, the concept of "Full Impedance compensation" applied to 
>> >power distribution system with decoupling capacitors sounds an interest 
>> >concept to achieve the target impedance over the wide frequency range of a 
>> >power distribution network with a damped or fully compensated impedance 
>> >which eliminates the anti-resonant peaks due to parallel RLC tank formed by 
>> >the decoupling capacitors.
>>
>>  
>>
>> I would like your guidance on recommended source of information about the 
>> concept of Full Impedance Compensation applied on Power Distribution System 
>> for the reduction of the impedance (damped anti-resonant impedance). 
>>
>>  
>>
>> - Has this concept being applied to real applications?
>>
>> - Is this concept used on comercially available SI simulation tools?
>>
>>  
>>
>> Thanks.
>>
>>  
>>
>> Rodrigo
>>
>>     
------------------------------------------------------------------
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:
http://www.freelists.org/webpage/si-list

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


List technical documents are available at:
                http://www.si-list.net

List archives are viewable at:     
                http://www.freelists.org/archives/si-list
 
Old (prior to June 6, 2001) list archives are viewable at:
                http://www.qsl.net/wb6tpu

Other related posts: