[ibis-macro] Re: input to Tx analog model
- From: "Muranyi, Arpad" <Arpad_Muranyi@xxxxxxxxxx>
- To: <ibis-macro@xxxxxxxxxxxxx>
- Date: Thu, 20 Jan 2011 16:53:32 -0800
I got stuck in my thinking on one detail question:
If the Tx GetWave output is a difference waveform but
the analog Tx model has a differential in/out (4-port),
how do we convert the difference waveform coming from
Tx GetWave into a differential waveform for the Tx
analog model?
Since the input to Tx GetWave is +/- 0.5 volt, centered
around 0, it would be tempting to say, let's just make
a copy of the Tx GetWave output, multiplied by -1. But
this only works if the difference waveform from Tx GetWave
is guaranteed to be DC free. I am not sure if that is a
safe assumption.
Does anyone have any suggestions?
Arpad
===========================================================
From: fangyi_rao@xxxxxxxxxxx [mailto:fangyi_rao@xxxxxxxxxxx]
Sent: Thursday, January 20, 2011 4:33 PM
To: Muranyi, Arpad; ibis-macro@xxxxxxxxxxxxx
Subject: RE: [ibis-macro] Re: input to Tx analog model
Arpad;
Yes, we are asking the same questions. While yours are more focusing on
AMI simulation, mine are focusing on general SPICE transient simulation.
A subtitle difference is that in AMI it's the EDA tool that internally
create the input to Tx analog models but in transient user has to create
the input explicitly on the schematic. So these questions are more
prominent in transient simulations.
Regards,
Fangyi
From: ibis-macro-bounce@xxxxxxxxxxxxx
[mailto:ibis-macro-bounce@xxxxxxxxxxxxx] On Behalf Of Muranyi, Arpad
Sent: Thursday, January 20, 2011 2:16 PM
To: ibis-macro@xxxxxxxxxxxxx
Subject: [ibis-macro] Re: input to Tx analog model
Fangyi,
You are basically asking the same question to which I was
trying to find the answer in the last few ATM meetings and
email discussions on the topic of the AMI analog models.
The difference you describe is exactly why I am now asking
the questions "what is ***the*** analog model" or AMI? Is
the stimulus PWL source included or not? Do we need two
different types of models? Do we need to describe the
different usages and/or flows... I still do not see the end
of the tunnel in this discussion.
I thought when Walter posted his HSPICE netlist
http://www.vhdl.org/pub/ibis/macromodel_wip/archive/20101214/walterkatz/
BIRD%20122%20Sample%20SPICE%20Deck/generic_tstonefile_spice.pdf
that we had the answer. But when Walter brought up the
possibility of using the output of Tx GetWave as the input
to the Tx analog model, I lost that feeling. I am still
looking for answers and suggestions on how to write the
specification in an unambiguous way to describe the analog
models.
I am open to any suggestions...
Thanks,
Arpad
============================================================
From: ibis-macro-bounce@xxxxxxxxxxxxx
[mailto:ibis-macro-bounce@xxxxxxxxxxxxx] On Behalf Of
fangyi_rao@xxxxxxxxxxx
Sent: Wednesday, January 19, 2011 9:11 PM
To: ibis-macro@xxxxxxxxxxxxx
Subject: [ibis-macro] input to Tx analog model
Hi, Experts,
In light of IBIS analog model enhancement discussed at recent ATM
meetings, I want to raise a question: in SPICE level transient
simulations (w/o AMI), what is the input signal to the Tx analog model?
In IBIS 5.0 the answer is simple. The input is digital signal whose
threshold crossing events trigger rise and fall edges being formed and
transmitted. The swing and waveform of the edges are part of the analog
model. They are independent of the input signal and user doesn't need to
know about them.
If we are to make Tx analog model able to accept AMI_GetWave output
waveform as input, we need to abandon this notion of event trigger and
make the input signal a continuous analog waveform (e.g. PWL). The
output waveform of Tx analog model will be directly shaped by the input
waveform. As a result, to run a SPICE level transient simulation (w/o
AMI) user has to look up the swing and slope parameters (Voh, Vol,
trise, etc) from model files in order to create the input signal with
proper amplitude and shape.
Is this an acceptable use model for transient? If user replace his old
model with a new one, then the result is most likely wrong because his
old digital input signal doesn't have the correct amplitude and slope.
Regards,
Fangyi
Other related posts:
