[SI-LIST] Re: Accounting random jitter

Instead on focusing on SPICE ability to incorporate random jitter into
an eye diagram it is fundamentally more important to determine what
relevant analysis is required.  For example, if you want to simulate the
integrity of a full backplane, including the RJ generated by an ASIC
SERDES, SPICE may not be the right tool.  (BTW, we are getting some
marvelous correspondence between the BERT-DSO measurements and IConnect
3.2.1 version for examining DJ degradation in eye diagrams)

Scott is right, SPICE is not a stochastic simulation tool in the
strictest sense.  Routinely eye diagrams are examined consisting of ISI,
DCD, even SJ (periodic) and this analysis works well since the total
deterministic jitter can be represented by a bounded probability density
function - in other words the peak-peak jitter describes the process.
Now, add RJ jitter as a degradation to the eye diagram (assuming your
simulator, GUI, etc., supports it).   You now have the convolution
issues with adding RJ and DJ (Ransom Stephens has written some good app
notes on this subject btw), but more importantly you have the issue of
stochastic significance since RJ is unbounded and requires many bit
periods.  10^-12 BER is 1 bit failure in 10^12 bits and you'll need a
min of 30 bit errors for statistical significance.  Frankly, I doubt a
time-step type of simulator like SPICE is the right tool to deal with
this, and even with the ability to add time domain jitter does it
accomplish a valid new analysis or simulation method?  Would it help in
understanding the integrity of the link, in other words.

This would make a great DesignCon Techforum or panel discussion for the
likes of Ken Kundert, Mike Li, and Ransom Stephens!  







Alfred P. Neves      <*)))))><{
Teraspeed Consulting Group LLC 
121 North River Drive 
Narragansett, RI 02882
 
Hillsboro Office
735 SE 16th Ave.
Hillsboro, OR, 97123
(503) 679 2429 Voice
(503) 210 7727 Fax
 
Main office
(401) 284-1827 Business
(401) 284-1840 Fax 
http://www.teraspeed.com
 
Teraspeed is the registered service mark 
of Teraspeed Consulting Group LLC
 


-----Original Message-----
From: Fred Balistreri [mailto:fred@xxxxxxxxxxxxx] 
Sent: Tuesday, August 23, 2005 2:47 PM
To: scott@xxxxxxxxxxxxx; Pratt, Gary
Cc: ray.anderson@xxxxxxxxxx; james.plesa@xxxxxxx; edpc108@xxxxxxxxx;
si-list@xxxxxxxxxxxxx; ransom@xxxxxxxxxxxxx; al@xxxxxxxxxxxxx
Subject: Re: [SI-LIST] Re: Accounting random jitter


There seems to be a disconnect between "random noise generation" and
simulation of jitter. So let me just say that random noise generation is
not only possible in SPICE but it has been done. It can be built into
the code or a random number generator wrt. time can be used and then
incorporated with the SPICE PWL voltage, current statement. This will
give you random noise voltage or current. However jitter implies you
have a semiconductor model and the noise needs to be built into it so
that is matches in some way to measurement.  There are various sources
for jitter in semiconductors. But the majority of jitter comes from
power supply / gnd noise.  Variation of the eye due to transmisison line
effects is not in my opinion clasical jitter although it can be
characterized (measured) as such.  The rest of classical jitter comes
from the aforementioned junction noise, johnson, 1/f and thermal. The
noise from power/gnd can be simulated by SPICE and that jitter shows up
in the time domain. Non-deterministic jitter as defined by Scott is a
much harder model to come up with. To date I only know of one EDA
company that claims to have a working mathematical model to simulate it.

My point is there is a lot of ways to introduce jitter. It can be done
in a behavioral model as well. But non-deterministic jitter of
semiconductors is not trivial and takes a bit more than just a random
noise generator. Once measured and known then its easy to incorporate
into SPICE, but that is after the fact.

Measuring or plotting the eye and post processing it has also been done
in SPICE without any external code. It's a function of the GUI. Be wise
when selecting your SPICE software.

Best Regards,
Fred

----- Original Message ----- 
From: "Scott McMorrow" <scott@xxxxxxxxxxxxx>
To: "Pratt, Gary" <gary_pratt@xxxxxxxxxx>
Cc: <ray.anderson@xxxxxxxxxx>; <james.plesa@xxxxxxx>;
<edpc108@xxxxxxxxx>; <si-list@xxxxxxxxxxxxx>; <ransom@xxxxxxxxxxxxx>;
<al@xxxxxxxxxxxxx>
Sent: Tuesday, August 23, 2005 12:56 PM
Subject: [SI-LIST] Re: Accounting random jitter


> All
> Let me elaborate.  I assumed that the question referred to true random

> unbounded jitter, as the term random jitter (Rj) implies, which has a 
> statistically infinite distribution.  Any deterministic simulator will

> necessarily bound the noise distribution over time and will result in 
> results that are generally better than actual measurements.  Of 
> course, if you are able to model a statistically unbounded gaussian 
> source and have sufficient compute time, you can use a deterministic 
> simulator to resolve down to whatever confidence you require.  But, 
> even measurement of Rj with instruments at a 10 Gbps data rate are 
> generally limited to rather high bit error rates (10^-12 BER), because

> of the huge number of bits that must be measured.
>
> Al or Ransom might want to chime in, and add their knowledge.
>
> Regards,
>
> Scott
>
> Scott McMorrow
> Teraspeed Consulting Group LLC
> 121 North River Drive
> Narragansett, RI 02882
> (401) 284-1827 Business
> (401) 284-1840 Fax
>
> http://www.teraspeed.com
>
> TeraspeedR is the registered service mark of
> Teraspeed Consulting Group LLC
>
>
>
> Pratt, Gary wrote:
>
> >Mentor's Eldo can use all the normal linear noise analysis device 
> >specifications (flicker, thermal, junction, etc) in a transient 
> >analysis as described by this excerpt from the Eldo manual:
> >
> >Transient noise simulation can be applied to all types of circuit 
> >without restriction. To perform transient noise analysis, physical 
> >noise of electrical devices is emulated by time dependent
> >current sources. The frequency characteristics of these sources are
> >referred to the noise models
> >of the noisy components. The method used is simple, fast and does not
> >disturb the simulated
> >behavior of the circuit because the noise signals introduced are
> >continuous and fully
> >deterministic.
> >
> >However last I'd heard, this was unique to Eldo with the other SPICE 
> >derivatives behaving as Ray described.
> >
> >One could also incorporate any type of desired time-domain noise 
> >effect in IBIS 4.1 AMS models.
> >
> >
> >Gary
> >
> >
> >
> >-----Original Message-----
> >From: si-list-bounce@xxxxxxxxxxxxx 
> >[mailto:si-list-bounce@xxxxxxxxxxxxx]
> >On Behalf Of Ray Anderson
> >Sent: Tuesday, August 23, 2005 1:34 PM
> >To: james.plesa@xxxxxxx; scott@xxxxxxxxxxxxx; edpc108@xxxxxxxxx
> >Cc: si-list@xxxxxxxxxxxxx; Ray Anderson
> >Subject: [SI-LIST] Re: Accounting random jitter
> >
> >I always thought the .noise function in spice merely determined the 
> >total RMS noise contribution of the various circuit components 
> >(resistive and active) whose noise performance could either be 
> >calculated or was specified. It isn't obvious to me that this 
> >function can be used to inject jitter into a time domain simulation.
> >
> >Can spice's Monte Carlo statistical capabilities be applied to the 
> >zero crossings of a voltage source? I've always thought that it could

> >only statistically vary component values.
> >
> >On the other hand one could probably generate a random noise voltage 
> >and sum it with a reference voltage in a comparator to cause random 
> >jitter on the output of the comparator. Alternatively, the random 
> >noise source could be used to modulate a behavioral VCO clock source.

> >This basically emulates some techniques that can be applied to 
> >physical test equipment in the lab.
> >
> >
> >-Ray
> >
> >
> >-----Original Message-----
> >From: si-list-bounce@xxxxxxxxxxxxx 
> >[mailto:si-list-bounce@xxxxxxxxxxxxx]
> >On Behalf Of Plesa, James T.
> >Sent: Tuesday, August 23, 2005 11:00 AM
> >To: scott@xxxxxxxxxxxxx; edpc108@xxxxxxxxx
> >Cc: si-list@xxxxxxxxxxxxx
> >Subject: [SI-LIST] Re: Accounting random jitter
> >
> >Most or all SPICE simulators that I have used have included .NOISE 
> >analysis and Monte-Carlo analysis, either of which would seem suited 
> >to introducing eye-diagram jitter.=3D20
> >
> >Jim Plesa
> >Northrop Grumman Corp.=3D20
> >
> >-----Original Message-----
> >From: si-list-bounce@xxxxxxxxxxxxx 
> >[mailto:si-list-bounce@xxxxxxxxxxxxx]
> >On Behalf Of Scott McMorrow
> >Sent: Tuesday, August 23, 2005 12:45 PM
> >To: edpc108@xxxxxxxxx
> >Cc: si-list@xxxxxxxxxxxxx
> >Subject: [SI-LIST] Re: Accounting random jitter
> >
> >Jayaprakash
> >
> >Unfortunately, no.  SPICE is a deterministic simulator and does not 
> >simulate stochasic (random) processes.  You will need something like 
> >Agilent ADS, Ansoft Designer, StatEye, or roll-your-own Matlab code 
> >to integrate random noise into your simulations.=3D20
> >
> >regards,
> >
> >Scott
> >
> >Scott McMorrow
> >Teraspeed Consulting Group LLC
> >121 North River Drive
> >Narragansett, RI 02882
> >(401) 284-1827 Business
> >(401) 284-1840 Fax
> >
> >http://www.teraspeed.com
> >
> >Teraspeed(r) is the registered service mark of Teraspeed Consulting 
> >Group LLC
> >
> >
> >
> >Jayaprakash wrote:
> >
> >
> >
> >>Hi,
> >>
> >> Is there a std. way to include/account random jitter effects in =
> >>
> >>
> >a=3D20
> >
> >
> >>typical eye diagram simulation in SPICE?
> >>
> >> Thanks and regards,
> >>    Jayaprakash.=3D20
> >>
> >>
> >> =3D09
> >>
> >>
> >
> >
> >
> >
>
>
> ------------------------------------------------------------------
> 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 FAQ wiki page is located at:
>                 http://si-list.org/wiki/wiki.pl?Si-List_FAQ
>
> List technical documents are available at:
>                 http://www.si-list.org
>
> List archives are viewable at: 
> http://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:
http://www.freelists.org/webpage/si-list

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

List FAQ wiki page is located at:
                http://si-list.org/wiki/wiki.pl?Si-List_FAQ

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

List archives are viewable at:     
                http://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: