Hi Arpad,
I don't see why there is a problem here. In both cases the threshold would be
0V. Please note that threshold is always used together with parameter
Rx_Sensitivity.
Regards,
Fangyi
From: ibis-macro-bounce@xxxxxxxxxxxxx <ibis-macro-bounce@xxxxxxxxxxxxx> On
Behalf Of Muranyi, Arpad
Sent: Tuesday, July 02, 2019 7:12 PM
To: ibis-macro@xxxxxxxxxxxxx
Subject: [ibis-macro] Re: Question on BIRD197.3
[EXTERNAL]
Walter, and All,
Here is a drawing to illustrate my comment about how we seem to be
misusing or abusing the word "threshold" in our discussions (and various
specifications).
[cid:image001.png@01D535E4.7AADE510]
There are two scenarios in this drawing. The one on the top is usually
considered single ended
signaling, and the one on the bottom is usually considered differential
signaling. Yet in both
cases, the receiving input buffer could be built with the same exact circuit on
the die. In the
first case, the DC source providing the reference may be on the die, or
externally provided
through a pin. Note that the signal in the top case is centered around that 2
V DC source. But
we can just as well provide that "reference voltage" as the inverted signal of
a differential pair,
as shown on the bottom plot. From the perspective of the receiving buffer it
does NOT matter
which configuration it is connected to. The "difference waveform" it sees is
the same in both
cases. The receiver only cares about when the two waveforms cross each other.
The only thing
we need to watch out for is not to exceed the common mode range this receiver
is able to
handle.
With all this in mind, I would be curious to find out what people think is the
meaning of
"threshold", especially in the case of differential signaling when we lost the
DC component
of the signal and are dealing with a difference waveform only. I suspect that
we have a
terminology problem here, and this is what makes me feel that we are abusing
this term.
I will stop right here, even though I have a lot more on my mind about all
this. I would like to
find out first what people would say is the meaning of threshold with the above
in mind...
Thanks,
Arpad
==========================================================================
From: ibis-macro-bounce@xxxxxxxxxxxxx<mailto:ibis-macro-bounce@xxxxxxxxxxxxx>
[mailto:ibis-macro-bounce@xxxxxxxxxxxxx] On Behalf Of Muranyi, Arpad
Sent: Tuesday, July 2, 2019 7:27 PM
To: ibis-macro@xxxxxxxxxxxxx<mailto:ibis-macro@xxxxxxxxxxxxx>
Subject: [ibis-macro] Re: Question on BIRD197.3
Walter,
That is the point I was trying to make, the Tx doesn't know any of that.
But if the definition of the channel impulse response is that it is taken
at the Rx die pad, and the Tx GetWave output convolved with the channel
IR is also considered to be the Rx die pad waveform, than the Tx output
should be centered around the x-axis. Consequently if the Tx had any
internal artifacts which would have added some DC component to the
waveform it received (which was centered on the x-axis), then the Tx
should remove that DC component from its output, and add that value
to DC_Offset, so in case someone wants to reproduce the real waveform
at the output of Rx GetWave, they would be able to "figure in" the amount
of DC shift that came from the Tx GetWave too...
Thanks,
Arpad
===========================================================
From: Walter Katz [mailto:wkatz@xxxxxxxxxx]
Sent: Tuesday, July 2, 2019 1:55 PM
To: Muranyi, Arpad <Arpad_Muranyi@xxxxxxxxxx<mailto:Arpad_Muranyi@xxxxxxxxxx>>;
ibis-macro@xxxxxxxxxxxxx<mailto:ibis-macro@xxxxxxxxxxxxx>
Subject: RE: [ibis-macro] Question on BIRD197.3
Arpad,
And how does the Tx GetWave know how to put in a DC Offset, or any offset at
all. All it can possibly have is the IR of the channel that was passed into Tx
AMI_Init. DC_Offset is not an input to the Tx model.
Walter
Walter Katz
wkatz@xxxxxxxxxx<mailto:wkatz@xxxxxxxxxx>
Office 978.461-0449 x 133
Mobile 720.417-3762
[cid:image002.jpg@01D47B36.D40C66E0]
From: ibis-macro-bounce@xxxxxxxxxxxxx<mailto:ibis-macro-bounce@xxxxxxxxxxxxx>
<ibis-macro-bounce@xxxxxxxxxxxxx<mailto:ibis-macro-bounce@xxxxxxxxxxxxx>> On
Behalf Of Muranyi, Arpad
Sent: Tuesday, July 2, 2019 2:30 PM
To: ibis-macro@xxxxxxxxxxxxx<mailto:ibis-macro@xxxxxxxxxxxxx>
Subject: [ibis-macro] Question on BIRD197.3
Hello Everyone,
I noticed something while reading the BIRD.
The definition makes the following statement:
"The mean value of the steady state high and low voltages of the channel at the
Rx pad."
Notice that this basically says that the DC_Offset is defined as the DC
component of the
waveform at the Rx die pad. This makes sense, because the channel impulse
response
we pass into the AMI simulation (Tx Init) is based on the waveform at the Rx
die pad.
Now, the problem I see is that in the GetWave flow the Rx GetWave function gets
the
output waveform of the Tx GetWave function (convolved with the channel IR). I
don't
see anything in the spec that Tx GetWave has to return a zero DC component
waveform.
In our discussions it was even said that the Tx GetWave is allowed to return a
waveform
with a DC component.
But ask ourselves the question: What does the waveform coming out of the Tx
GetWave
and convolved with the channel IR represent? As far as I can tell, that
waveform is the
equivalent of the Rx die pad waveform.
So I see a conflict here. The DC_Offset parameter defines the Rx die pad
waveform as
a zero DC component waveform, but at the same time we allow the Tx GetWave
function
to produce a waveform WITH a DC component. I think the definition of this
DC_Offset
parameter implies that the Tx GetWave output must have NO DC component, but we
don't say that anywhere.
Or is my thinking flawed somewhere?
Thanks,
Arpad
======================================================================