[SI-LIST] Re: Effective Radius of Decoupling Capacitor
- From: Jack Si <sij99@xxxxxxxxx>
- To: "istvan.novak@xxxxxxxxxxx" <istvan.novak@xxxxxxxxxxx>
- Date: Wed, 28 Aug 2013 06:26:42 -0700 (PDT)
Istavan,
Thanks for the valuable in-depth info.
Thanks,
Jack
________________________________
From: Istvan Novak <istvan.novak@xxxxxxxxxxx>
To: sij99@xxxxxxxxx
Cc: "si-list@xxxxxxxxxxxxx" <si-list@xxxxxxxxxxxxx>
Sent: Wednesday, August 28, 2013 6:18 PM
Subject: [SI-LIST] Re: Effective Radius of Decoupling Capacitor
Jack,
The short column cited in my previous message gives the expressions
people use in these arguments, but there are other important aspects
that the short column due to lack of space did not touch upon. None of
the usual arguments around service area seem to care about the question
whether the charge available within the service radius is enough for the
load. And an even more basic question: why do we care where the charge
comes from? Why do we want all of the charge for a load transient to
come from a capacitor at a specific location? Again, in the old days,
using very few capacitors and mostly traces for PDN, ignoring these
questions made a little more sense. But if you consider another
extreme, a magic transmission line, matched, with its characteristic
impedance matching the target impedance for the PDN, charge will flow
as the load demands and we never have to worry about where the charge is
coming from.
Regarding bulk capacitors, the same considerations apply, just the
numbers going into these considerations will be different based on the
frequency band we expect to cover with the bulk capacitors. Most of the
time this means that bulk capacitors have less restrictions in
placement, though when a large amount of current flows, we still need to
be careful not to let that current flow over large distances even if the
static and dynamic voltage drops otherwise would vbe acceptable on the
PDN, simply because traveling a larger distance the transient noise
current would then have more possibility to couple into a sensitive area.
Regards,
Istvan Novak
Oracle
On 8/27/2013 11:40 PM, Jack Si wrote:
> Hi Istvan,
> I put my question in two different views,
>
> 1. From the reference, I found that the service area or effective radius is
> calculated for the PCB by considering Dk and the rise time of the excitation.
> So I infer that the decoupling capacitor should be placed within the service
> area so the we can maintain the charge hold by the path. Isn't right?
>
> 2. The method ,I brought out taken the effective radius of the capacitor (may
> be decoupling or bulk), In case of bulk, how it should be distributed
> throughout the plane and for decoupling, how near it should be placed to the
> load for its effectiveness. Correct me if my inference is wrong.
>
> Thanks
> Jack
>
>
>
>
> ________________________________
> From: Istvan Novak <istvan.novak@xxxxxxxxxxx>
> To: sij99@xxxxxxxxx
> Cc: "si-list@xxxxxxxxxxxxx" <si-list@xxxxxxxxxxxxx>
> Sent: Tuesday, August 27, 2013 6:28 PM
> Subject: [SI-LIST] Re: Effective Radius of Decoupling Capacitor
>
>
> Jack,
>
> The often mis-used concept of service radius of capacitors dates back to
> the days when there were hardly any planes on the PDN, most chips were
> fed with wide traces and capacitors were placed as close to the device
> pins as possible. To answer the question "How close the capacitor
> should be?" under these circumstances, the service area concept might
> give some useful insight. With planes and multiple capacitors I find
> this concept often very misleading.
>
> It might be more useful to think in terms of extreme-terminated
> transmission lines. Imagine that you have a dead short at the location
> of the capacitor and you connect this short to the observation point
> (pin of chip) with a transmission line, formed by the trace or the plane
> shape making the connection. The pin of the chip will see the input
> impedance of the shorted transmission line, which, if the line is
> electrically shorter than quarter wave (or odd-integer multiples, but we
> dont want to go there unless our PDN has to service known comb-line
> noise spectrum), will be inductive. The inductive input impedance is
> proportional to the electrical distance through the tangent function, so
> the farther we place the capacitor, the higher the inductance becomes.
> Dependent on where you may loose effectiveness of bypassing, you can
> then draw a conclusion of the 'service area' or how far you can place
> the capacitor to still do something useful. Instead of using ideal
> short, you can also modify the transmission line expressions to include
> the realistic impedance of the capacitor.
>
> Though it discusses laminates, the same reasoning can be found in more
> detail in
> http://www.electrical-integrity.com/Quietpower_files/Quietpower-16.pdf
>
> Regards,
>
> Istvan Novak
> Oracle
>
>
>
> On 8/27/2013 3:52 AM, Jack Si wrote:
>> Hi experts,
>> I read from an application note that the effective radius of the capacitance
>> if 0.005*lamda. lamda is the actual wavelength of the capacitor's resonance
>> frequency.i.e, 2pi*vp*sqrt(LC). where vp is the propagation velocity. From
>> this equation, i infer that the inductance(ESL) and capacitance are directly
>> proportional in square root. i.e, radius increase with the increase of ESL
>> or C.
>>
>>
>> But in paper "Effective Decoupling Radius of Decoupling Capacitor" i found
>> it is inversely proportional. Please suggest me where i miss the way.
>>
>>
>> Thanks and Regards,
>>
>> Jack
>>
>>
>
------------------------------------------------------------------
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 forum is accessible at:
http://tech.groups.yahoo.com/group/si-list
List archives are viewable at:
//www.freelists.org/archives/si-list
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 forum is accessible at:
http://tech.groups.yahoo.com/group/si-list
List archives are viewable at:
//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:
