[SI-LIST] Re: AC series capacitor position in high speeddifferential signals
- From: Istvan Novak <istvan.novak@xxxxxxx>
- To: "si-list@xxxxxxxxxxxxx" <si-list@xxxxxxxxxxxxx>
- Date: Wed, 19 May 2010 00:03:49 -0400
Just an FYI: off-line we followed up with Yuriy to see where our
differences were coming from.
It turns out that it was in the different ways how we can define a
channel and a system. Yuriy
uses the definition which assumes that the source and load impedances
move with the ports as
we test for reciprocity. In my nomenclature I am more focusing on what
we may call the channel,
which does not include the source and load, and consider the impact of
different source and load
impedances in a separate concatenation step. Bottom line: Lorentz
theorem of course holds, this
however means that for the reciprocity test we move the source and load
impedances as well.
If the source and load impedances stay put, and we just move around the
DC blocking capacitor, the
voltage transfer function as well as the resulting received eye can (and
most of the time do) change.
The degree of change, of course, depends on several things: the severity
of discontinuities around
the DC blocking capacitor and the total losses through the channel.
Well-designed DC-blocking
capacitors, as Yuriy's slides illustrate it very nicely, can result in
very minimal disturbance. Low-loss
channels, on the other hand, will make even lower discontinuities more
'visible'.
Regards,
Istvan Novak
Oracle-America
Istvan Novak wrote:
> Chris,
>
> Any combination of linear resistors and linear reactances would still be
> linear. Any voltage
> dependence in any of them would make the combination nonlinear. Though
> when I mentioned
> receiver nonlinearity in my earlier posting, I had more drastic versions
> in mind, when
> something gets driven into saturation.
>
> Last night I sent an off-line message to Yuriy to clarify further what I
> meant by transfer
> function...
>
>
> Best Regards,
>
> Istvan Novak
> Oracle-America
>
>
>
> Chris Cheng wrote:
>
>> Yuriy,
>> Do we still consider the end points linear if the transeivers are
>> combination of parasitics capactiors (due to ESD structures) and resistive
>> terminator (tx and rx circuits) ? In particular, the ESD capacitance is
>> voltage dependent ?
>> Thanks in advanced,
>> Chris
>>
>> ________________________________________
>> From: si-list-bounce@xxxxxxxxxxxxx [si-list-bounce@xxxxxxxxxxxxx] On Behalf
>> Of Yuriy Shlepnev [shlepnev@xxxxxxxxxxxxx]
>> Sent: Saturday, May 15, 2010 1:15 PM
>> To: 'Istvan Novak'
>> Cc: leeritchey@xxxxxxxxxxxxx; 'fei xue'; si-list@xxxxxxxxxxxxx
>> Subject: [SI-LIST] Re: AC series capacitor position in high
>> speeddifferential signals
>>
>> Istvan,
>>
>> Lorentz theorem states that "the response of a system constructed of linear
>> reciprocal materials to a source is unchanged when source and measurer are
>> interchanged". It means that the transfer function from a source with 75-Ohm
>> terminator to a 25-Ohm load for instance is exactly the same as the transfer
>> in the opposite direction. I guess you meant that the transfer function will
>> be different if you swap the termination impedances - it is so, but still
>> identical in both directions. It might be counterintuitive and if the
>> differences in the transfer functions or in the transmission parameters are
>> observed - for PCB applications it means either non-linearity of some
>> elements or errors in measurements or computations (as far as I know
>> non-reciprocal materials such as biased ferrites are not used for PCBs and
>> packaging).
>> I have added a few more illustration for a case of mismatched source and
>> load and for a case with asymmetric resonances to the app note at
>> http://www.simberian.com/AppNotes/AC_CouplingCapacitors_2010_02.pdf (see the
>> backup slides).
>>
>> Best regards,
>> Yuriy
>> www.simberian.com
>>
>>
>>
>> -----Original Message-----
>> From: si-list-bounce@xxxxxxxxxxxxx [mailto:si-list-bounce@xxxxxxxxxxxxx] On
>> Behalf Of Istvan Novak
>> Sent: Friday, May 14, 2010 9:16 PM
>> To: Yuriy Shlepnev
>> Cc: leeritchey@xxxxxxxxxxxxx; 'fei xue'; si-list@xxxxxxxxxxxxx
>> Subject: [SI-LIST] Re: AC series capacitor position in high
>> speeddifferential signals
>>
>> Yuriy,
>>
>> Nice app note, this will be useful for people to comprehend reciprocity
>> and symmetry. With Simbeor
>> one can also model the via transitions along the path. The slides
>> nicely illustrate the point that in a linear
>> but asymmetrical channel the transfer parameters are the same from
>> either end IF there are no further
>> reflections. However, if in a next step you also simulate the
>> source-to-load voltage transfer function with
>> slightly mismatched transmit and receive impedances, the transfer
>> function and the eye will be direction
>> dependent even if everything is linear.
>>
>> Best regards,
>> Istvan
>>
>>
>>
>>
>> Yuriy Shlepnev wrote:
>>
>>
>>> Istvan and Lee,
>>> I guess the next step in resolving the contradiction would be validation
>>>
>>>
>> of
>>
>>
>>> your measurements with a simulation that captures all relevant details of
>>> the problem :)
>>> For what it worth, I simulated the effect of the capacitor placement in a
>>> simple channel - the results and some observations on the subject are in
>>>
>>>
>> App
>>
>>
>>> Note #2010_02 at http://www.simberian.com/AppNotes.php. As expected,
>>> swapping ideal driver and receiver does not change the transmission at all
>>> due to the reciprocity principle (the statement is valid for any linear
>>> reciprocal system). Though the reflection strongly depends on the location
>>> of the capacitor and may affect the overall signal transmission due to
>>> non-linearity and reflections from the driver and receiver.
>>>
>>> Best regards,
>>> Yuriy Shlepnev
>>> www.simberian.com
>>>
>>> -----Original Message-----
>>> From: si-list-bounce@xxxxxxxxxxxxx [mailto:si-list-bounce@xxxxxxxxxxxxx]
>>>
>>>
>> On
>>
>>
>>> Behalf Of Istvan Novak
>>> Sent: Thursday, May 13, 2010 6:02 PM
>>> To: leeritchey@xxxxxxxxxxxxx
>>> Cc: fei xue; si-list@xxxxxxxxxxxxx
>>> Subject: [SI-LIST] Re: AC series capacitor position in high
>>> speeddifferential signals
>>>
>>> Lee,
>>>
>>> I agree with you that people should not blindly trust only simulation
>>> results. Those who follow my work
>>> know that for me a question answered means that measurements,
>>> simulations and understanding all have
>>> to agree to a sufficient degree. That having said I cant argue with
>>> your statement when you say that your
>>> measurements show negligible reflections in your boards: I trust it is
>>> true for the boards you built.
>>>
>>> What prompted my posting was that your original comment could be
>>> interpreted as a suggestion that
>>> the capacitor placement makes no difference under any circumstances as
>>> long as the channel is linear.
>>> My point was that there are situations (and we dont need to think about
>>> something exotic, just cases
>>> when pad discontinuity and/or via stub discontinuities are not removed
>>> for any reason) when the location
>>> of capacitor makes a difference in the received eye parameters by a
>>> noticeable way. With very typical
>>> parameters we can end up having 15-20% fluctuation of eye closure
>>> horizontally and vertically as we
>>> slide the capacitor along the channel. This can be simulated, but can
>>> also be measured on real system
>>> channels.
>>>
>>> Regards,
>>>
>>> Istvan Novak
>>> Oracle
>>>
>>>
>>>
>>> Lee Ritchey wrote:
>>>
>>>
>>>
>>>> I would maintain from my measurements that the reflections are negligibly
>>>> small.
>>>>
>>>> I would also maintain that simulation results, no matter what the tool or
>>>> operator, should not be trusted until they are validated by measurements
>>>>
>>>>
>>>>
>>> to
>>>
>>>
>>>
>>>> insure the modelling is accurate. I've seen too many simulations done
>>>> incorrectly which were used to develop design rules that were either
>>>> ineffective or caused problems that I, for one, will not use the results
>>>>
>>>>
>>>>
>>> of
>>>
>>>
>>>
>>>> unvalidated simulations.
>>>>
>>>> Each engineer can make up his or her own mind on this, but my experience
>>>> says validate simulation results before betting any money on them.
>>>>
>>>> I'm not trying to offend anyone who does simulations, just advising
>>>>
>>>>
>>>>
>>> proceed
>>>
>>>
>>>
>>>> with caution.
>>>>
>>>> Lee
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>> [Original Message]
>>>>> From: Istvan Novak <istvan.novak@xxxxxxx>
>>>>> To: Lee Ritchey <leeritchey@xxxxxxxxxxxxx>
>>>>> Cc: fei xue <harrison_cls@xxxxxxxxxxxx>; <si-list@xxxxxxxxxxxxx>
>>>>> Date: 5/12/2010 8:55:32 PM
>>>>> Subject: Re: [SI-LIST] Re: AC series capacitor position in high
>>>>>
>>>>>
>>>>>
>>>>>
>>>> speeddifferential signals
>>>>
>>>>
>>>>
>>>>
>>>>> Lee,
>>>>>
>>>>> As it was pointed out in earlier threads, location does make a
>>>>> difference unless reflections are
>>>>> negligibly small. Assuming linearity, when we move components around,
>>>>> reciprocity prevails,
>>>>> but voltage transfer function from source to load will change, which in
>>>>> turn impacts eye
>>>>> parameters.
>>>>>
>>>>> Regards,
>>>>>
>>>>> Istvan Novak
>>>>> Oracle-America
>>>>>
>>>>>
>>>>> Lee Ritchey wrote:
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>> This has been answered before on this forum. Since the circuits are
>>>>>> linear, it does not matter from a signal integrity point of view.
>>>>>>
>>>>>> Lee
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>> [Original Message]
>>>>>>> From: fei xue <harrison_cls@xxxxxxxxxxxx>
>>>>>>> To: <si-list@xxxxxxxxxxxxx>
>>>>>>> Date: 5/12/2010 11:35:49 AM
>>>>>>> Subject: [SI-LIST] AC series capacitor position in high speed
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>> differential signals
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>> Hello all,
>>>>>>> We often can get different guideline of placing capacitor position
>>>>>>>
>>>>>>>
>> when
>>
>>
>>>>>>>
>>>>>>>
>>>>>> placing AC series capacitor on high speed differential signals, like
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>> PCIe,
>>>>
>>>>
>>>>
>>>>
>>>>>> SAS or LVDS signals. sometimes we followed the guideline to put
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>> capacitors
>>>>
>>>>
>>>>
>>>>
>>>>>> near driver, sometimes near multi-connection connectors or sometimes
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>> put it
>>>>
>>>>
>>>>
>>>>
>>>>>> near receivers.
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>> Could anybody tell me what is the consideration of capacitor placing
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>> position? Thanks!
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>> Harrison
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>
>>>>
>>>
>>>
------------------------------------------------------------------
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 technical documents are available at:
http://www.si-list.net
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:
