[ibis-macro] Re: redriver in spice simulation

  • From: <fangyi_rao@xxxxxxxxxxx>
  • To: <Arpad_Muranyi@xxxxxxxxxx>, <ibis-macro@xxxxxxxxxxxxx>
  • Date: Thu, 14 Mar 2013 11:21:02 -0600

Hi, Arpad;

I think some of your confusion arises from the wrong redriver AMI flow on the 
1st page in your drawing. The correct one, as plotted in my BIRD, is

Rx anlg -> Rx GetWave -> Tx GetWave -> Tx anlg

So the redriver simulation is just two back-to-back regular AMI simulations. 
The only thing special about redriver is that Rx GetWave output is the input to 
Tx GetWave.

Regarding your questions,


1)  how are the analog models used in legacy simulations
FR: My BIRD only concerns with AMI usage, and I don’t think it is meaningful to 
run SPICE simulation on redriver model w/o involving its algorithmic parts. So 
I tend to agree to Radek’s suggestion, namely ignoring the connection between 
the Rx half and Tx half in SPICE/non-AMI. But that’s beyond the scope of my 
BIRD.



2)  how are the analog models used in the IBIS-AMI simulations for
   channel characterization
FR: Given the correct flow described in the beginning of this email, it should 
be clear that in channel characterization, analog model is used in the exact 
same way as in a regular AMI simulation.


3)  how are the analog models used in the AMI flow
FR: Given the correct flow described in the beginning of this email, it should 
be clear that in AMI flow, analog model is used in the exact same way as in a 
regular AMI simulation.


4)  what is the nature of the analog models in either of these flows
FR: Given the correct flow described in the beginning of this email, it should 
be clear that in the nature of analog model is exactly the same as it’s in 
regular AMI flows.

Hope this helps.

Regards,
Fangyi

From: ibis-macro-bounce@xxxxxxxxxxxxx [mailto:ibis-macro-bounce@xxxxxxxxxxxxx] 
On Behalf Of Muranyi, Arpad
Sent: Wednesday, March 13, 2013 11:35 AM
To: ibis-macro@xxxxxxxxxxxxx
Subject: [ibis-macro] Re: redriver in spice simulation

Fangyi,

I deleted the emails below after your first message to make it easier
to see what you were referring to.  I see what you mean now.

We need to distinguish between the AMI flow and the legacy flow.
Not only that, we also have to define the nature of the input and
output of the Tx and Rx analog models.  Our current specification
leaves a lot of room for differences of opinion and arguments over
that topic as we have experienced it not too long ago.

What you are referring to below is the AMI flow.  What I was referring
to was the legacy flow.

A question was raised in the emails after your original posting on what
the analog models should do in a legacy simulation environment.  Should
they be cascaded as repeaters also?  Several comments were made to that
question.  One was to ignore the fact that these analog buffer models
are part of a repeater.  My comment was that they could also be cascaded
if we defined in the spec how that is done.  I am trying to illustrate
that in the attached PDF file.

However, in addition to that, your AMI flow seems to assume that the
the output of the Rx analog model and the input to the Tx analog model
are analog.  This needs to be defined, otherwise different people will
have different assumptions and with the wrong assumptions one will not
be able to implement your repeater proposal correctly.

So I think your BIRD should define the following:


1)  how are the analog models used in legacy simulations

2)  how are the analog models used in the IBIS-AMI simulations for
   channel characterization

3)  how are the analog models used in the AMI flow

4)  what is the nature of the analog models in either of these flows

Thanks,

Arpad
=======================================================================


From: fangyi_rao@xxxxxxxxxxx<mailto:fangyi_rao@xxxxxxxxxxx> 
[mailto:fangyi_rao@xxxxxxxxxxx]
Sent: Wednesday, March 13, 2013 12:59 PM
To: Muranyi, Arpad; ibis-macro@xxxxxxxxxxxxx<mailto:ibis-macro@xxxxxxxxxxxxx>
Subject: RE: [ibis-macro] Re: redriver in spice simulation

Arpad;

In a redriver, the Tx half is driven by the Rx half’s output signal and 
retransmits it to the downstream channel.

Fangyi


From: fangyi_rao@xxxxxxxxxxx<mailto:fangyi_rao@xxxxxxxxxxx> 
[mailto:fangyi_rao@xxxxxxxxxxx]
Sent: Tuesday, March 12, 2013 4:58 PM
To: Ambrish Varma; ibis-macro@xxxxxxxxxxxxx<mailto:ibis-macro@xxxxxxxxxxxxx>
Subject: redriver in spice simulation

Hi, Ambrish;

I am not sure if your suggestion of shorting redriver Rx ibis output pin to the 
redriver Tx ibis input pin in spice simulations will work because in spice 
simulations


1.       legacy Tx ibis model is driven by input signal threshold crossing 
events. That’s not how a redriver Tx half is driven by input signal.

2.       legacy Rx ibis model output is a digital signal of 1’s and 0’s 
(hopefully I am correct here). That’s not what redriver Rx half outputs.

Due to the uncertainties in redriver model behavior in non-AMI simulations, I 
prefer to keep my redriver BIRD separate from Walter’s repeater BIRD, at least 
for now.

Regards,
Fangyi


Other related posts: