[ibis-macro] Re: A rephrasing of my previous question, re: your GetWave() approach.
- From: "Muranyi, Arpad" <Arpad_Muranyi@xxxxxxxxxx>
- To: "ibis-macro@xxxxxxxxxxxxx" <ibis-macro@xxxxxxxxxxxxx>
- Date: Thu, 10 Jan 2013 21:01:14 +0000
Mike, Dave,
The statement: "the presumed physical location of the output needs to be the
same for Init() and GetWave()."
is incorrect, as far as I understand. But in order to do better
justice to this question, we need to be more specific about which
Init and GetWave function we are talking about (Tx or Rx).
Tx Init:
Input is the IR of the channel.
Output is the modified IR (if the function modifies it). The modification
is representative of the filtering activities in the Tx if there are any.
Since the function modifies the IR of the channel the result corresponds
to the waveform at the Rx input pad.
Tx GetWave:
Input is a digital time domain waveform, representing a bit pattern.
Output is a modified bit pattern whose amplitude is changed by the
filtering activities in the Tx. Since this does not involve the IR
of the channel, the output waveform does NOT correspond to the waveform
at the Rx pad, instead, it corresponds to the input to the analog
model of the channel (which could be seen as the Tx analog model's
input, according to the "alternate AMI flow" we discussed recently).
This is why I disagree with Mike's statement above. The output of
Tx Init and Tx GetWave do not correspond to the same physical location.
Thanks,
Arpad
==========================================================================
From: ibis-macro-bounce@xxxxxxxxxxxxx [mailto:ibis-macro-bounce@xxxxxxxxxxxxx]
On Behalf Of Mike Steinberger
Sent: Thursday, January 10, 2013 1:50 PM
To: ibis-macro@xxxxxxxxxxxxx
Subject: [ibis-macro] Re: A rephrasing of my previous question, re: your
GetWave() approach.
Dave-
At the very considerable risk of jumping into the middle of a conversation I
haven't been a part of, I suggest that for a given simulation, the presumed
physical location of the output needs to be the same for Init() and GetWave().
1. If the presumed physical location of the output is the input to the Rx Eq,
then however the EDA tool does its business, the net result should be the input
to the Rx Eq. for both statistical analysis and time domain simulation. This
choice is only valid if the Tx model is linear, in which case there are
several options open to the EDA tool developer, all of which are mathematically
equivalent.
2. If the presumed physical location of the output is the input to the analog
channel, then however the EDA tool does its business the net result should be
the input to the analog channel for both statistical analysis and time domain
simulation. This choice is valid for both linear or nonlinear Tx models. <This
is the cue for the usual discussion about interactions between a nonlinear
driver and a linear channel. Could we forgo that conversation just this once?>
The choices open to the EDA tool developer are a bit more limited in this case,
but there are definitely valid options available.
The EDA tool should be consistent about which of these two options it chooses
for a given simulation. Most Tx models will support either choice, in which
case the EDA tool is free to choose. There are, however, some models out there,
both Tx and Rx, which are only valid for one or the other of these two options.
Therefore, the EDA tool has to make a choice which is consistent with the
models it's running.
This is admittedly a very academic approach to your question because I don't
want to go into specifics any more than anyone else does.
Hope this helps.
Mike Steinberger
On 01/10/2013 01:23 PM, David Banas wrote:
Hi all,
At the end of yesterday's meeting, I asked the EDA vendors in the group if
they'd describe their GetWave() flows, for Tx modeling, to the group.
This request drew quite a bit of reluctance, understandably. I would like to
rephrase my question:
In the case of Tx modeling only, what processing flexibility do you gain by not
being required to accept the channel impulse response as input to the GetWave()
function?
My hope is that the above question is safer to answer and, yet, might still
further our common understanding of the reasons for the different assumptions,
regarding input signal and output location, made for Init() and GetWave(). If
you'll recall, after our recent re-read of Sec. 10 of IBIS v5.1 (and if we
agree to take Sec. 10 as a normative part of the standard, as opposed to just
one example interpretation of it), those assumptions are:
Flow Type
Input Signal
Presumed Physical Location of Output
Init()
Channel Impulse Response
Input to Rx Eq.
GetWave()
Binary Data Stream
Input to analog channel
If I am merely the last person to understand the motivation for the different
assumptions, above, I apologize for my obtuseness and thank you, in advance,
for your indulgence.
Best regards,
David Banas
Sr. Member Technical Staff
Altera<http://www.altera.com/>
+1-408-544-7667 - desk
Did you know Altera offers over 150 free online technical training
courses<http://www.altera.com/servlets/searchcourse?coursetype=Online&WT.mc_id=t9_ot_mi_mi_tx_a_311>?
Take one today!
________________________________
Confidentiality Notice.
This message may contain information that is confidential or otherwise
protected from disclosure. If you are not the intended recipient, you are
hereby notified that any use, disclosure, dissemination, distribution, or
copying of this message, or any attachments, is strictly prohibited. If you
have received this message in error, please advise the sender by reply e-mail,
and delete the message and any attachments. Thank you.
Other related posts:
