[ibis-macro] Re: BIRD 123.2 Intended Usage
- From: Mike Steinberger <msteinb@xxxxxxxxxx>
- To: ibis-macro@xxxxxxxxxxxxx
- Date: Wed, 25 Apr 2012 16:11:03 -0500
Greg-
Jitter modeling is more or less the same for most SerDes macros, and
this modeling can be performed outside the IBIS-AMI model itself; so it
makes sense to make jitter modeling a service of the EDA tool rather
than having a separate implementation in each IBIS-AMI model. The EDA
tool developer can implement these services once, and the model
developer only has to supply a few parameters. That's less work overall.
Furthermore, if the jitter modeling is performed by the EDA tool, it
will be performed the same way for all models, so the user isn't left
trying to figure out if one model is doing its jitter modeling
differently from another model.
For example, everyone thinks they know what RJ is, so we all believe we
mean the same thing when we refer to "RJ". The only proprietary aspect
involved is the actual level of the RJ, and that value usually ends up
being published for standards compliance anyway. So there's nothing left
that needs to be treated as proprietary inside a model.
The closest I can come to a problem that requires services inside the
IBIS-AMI model is the modeling of reference clock jitter as specified by
the Rx_Reference_Clock parameter. Even in this case, all that is
required of the model is that it includes the reference clock phase
noise in the operation of the the clock recovery loop. Our
implementation of that feature only took a couple of lines of code.
Other than that, I don't know of any jitter modeling analysis that
requires services in the IBIS-AMI model. Therefore, implementing jitter
modeling inside an IBIS-AMI model serves no purpose other than to
demonstrate the prowess of the model developer.
Hope this helps.
Mike S.
On 04/25/2012 03:26 PM, Gregory R Edlund wrote:
I'm addressing this question to the authors of BIRD 123.2, Walter,
Mike, and Todd.
As I understand this BIRD, the model developer has two choices for how
to model jitter: 1) inside the DLL or 2) outside the DLL using the
BIRD 123.2 parameters. What factors might influence a model developer
to choose option 1 over 2? Can someone give me an example?
Greg Edlund
Senior Engineer
Signal Integrity and System Timing
IBM Systems & Technology Group
3605 Hwy. 52 N Bldg 050-3
Rochester, MN 55901
Other related posts:
