[ibis-macro] Should a Init return a modified impulse response if GetWave_Exits is True?
- From: "Walter Katz" <wkatz@xxxxxxxxxx>
- To: "IBIS-ATM" <ibis-macro@xxxxxxxxxxxxx>
- Date: Mon, 12 Oct 2009 09:55:07 -0400
All,
Two years ago, we were correct in making the calling arguments to an Tx AMI
model and an Rx AMI model the same ?C? signature.
We made a mistake in trying to make the flows that a Tx and Rx model support
the same. The rules about Init_Returns_Impulse and GetWave_Exists are a
prime example.
For an Rx model it is perfectly understandable that a model writer does not
want to return an impulse response if the Rx equalization is not LTI. By
delivering a model that Init does not return an impulse response he is
forcing the EDA tool to do pattern dependant time domain analysis. The only
bad thing that happens is that the model developer forces the EDA tool to
spend additional time to get more accurate results. This exists today in
vendor models, and will continue to exist, and needs to be supported.
The situation for a Tx model is totally different. Today?s and foreseeable
Tx silicon, although sometimes having important non-LTI components, are
still well represented by an LTI equalization approximation. Tx models with
DCD being an example. It therefore should either be required, or highly
recommended that the Init function for Tx models do return a modified
impulse response when the TX GetWave exists to properly support Rx models
that require an best estimate representation of both the channel and the Tx
equalization.
Similarly, when I suggest that a model optionally return the impulse
response of just the filter, it is much more important to do so for the Rx
than the Tx, because the deconvolution step required in two of the flows
only applies to doing the deconvolution of the Rx Init output.
Walter
Walter Katz
303.449-2308
Mobile 720.333-1107
wkatz@xxxxxxxxxx
www.sisoft.com
Other related posts:
- » [ibis-macro] Should a Init return a modified impulse response if GetWave_Exits is True? - Walter Katz
