These references apply to linear time invariant single input - single output
systems. In particular, they assume that if the response on stimulus a(t) is
b(t), then response on K*a(t) is K*b(t). It is also assumed that this LTI
system is at rest (all voltages currents are zero) before the stimulus begins.
Let's take differential link with Tx and Rx buffers. Technically, the input is
a digital stimulus applied e.g. to the input gates of Tx buffer with different
polarities. It's not single input. The response caused by it is not scalable:
by doubling gate voltage, we don't get the output voltage doubled. The system
doesn't stay at rest either because we always have voltage at power terminals.
One possible solution is to define transition waveform first, for example, as a
voltage at Rx input when Tx switches from state low to high (or inverse swapped
horizontally). If we *agree* that this transition can be called "step response"
(although it doesn't meet the definition of step response), then impulse
response could be defined as its time derivative. However, someone else can say
that this waveform describes transition from state "low" to state "high"
instead of going from "neutral" to "high" and is a "doubled step response".
Hence, impulse response should be half of the derivative of this transition
I believe that what we mean by impulse response is a time derivative of the
transition waveform measured at Rx input, i.e. the first version, but still not
be sure that all IC vendors/AMI model developers mean the same. It would be
useful to provide this definition in the document. Of course, if we all agree
From: ibis-macro-bounce@xxxxxxxxxxxxx [mailto:ibis-macro-bounce@xxxxxxxxxxxxx]
On Behalf Of Walter Katz
Sent: Wednesday, October 25, 2017 7:51 PM
To: Muranyi, Arpad <Arpad_Muranyi@xxxxxxxxxx>; ibis-macro@xxxxxxxxxxxxx
Subject: [ibis-macro] Re: Question about "impulse response"
The impulse response of a channel has a clear definition in the industry. Why
does IBIS need one?
978.461-0449 x 133
[mailto:ibis-macro-bounce@xxxxxxxxxxxxx] On Behalf Of Muranyi, Arpad
Sent: Wednesday, October 25, 2017 9:01 PM
Subject: [ibis-macro] Question about "impulse response"
I would like to find out what the definition of "impulse response" is in
the AMI specification. I thought we had something about that in the
spec, but I can't find anything now that I am looking for it. Am I just
not seeing the forest from the trees, or is it really not defined?
I thought we did have something along the lines of "derivative of a
step response", but I can't find anything...