[ibis-macro] Re: Question about "impulse response"

  • From: Mike Steinberger <msteinb@xxxxxxxxxx>
  • To: ibis-macro@xxxxxxxxxxxxx
  • Date: Mon, 6 Nov 2017 11:31:29 -0600


Is the 800mV swing a single ended voltage swing or a differential voltage swing? The input signal to the receiver is a differential voltage. If we confuse single ended with differential, we'll be mixing apples and oranges, and end up with a dog's breakfast.

I cannot agree to Todd's interpretation until I get your answer to this question.

Mike Steinberger

On 11/6/2017 11:11 AM, Muranyi, Arpad wrote:


Thanks for that clarification.  Yes this is with the assumption that

the IR is derived from a step response, but all other methods

would need to provide equivalent results…

Would everyone on this email list agree with Todd’s interpretation?




*From:*Todd Westerhoff [mailto:twesterh@xxxxxxxxxx]
*Sent:* Monday, November 6, 2017 10:33 AM
*To:* Muranyi, Arpad <Arpad_Muranyi@xxxxxxxxxx>; ibis-macro@xxxxxxxxxxxxx
*Subject:* RE: [ibis-macro] Re: Question about "impulse response"


My interpretation is that the channel impulse response passed to Tx_Init will be the result of differentiating the 800 mV differential step response, as you say.

That, of course, is assuming that the impulse response is derived from a step response and not through some other method …


*Todd Westerhoff*

VP, Semiconductor Relations

Signal Integrity Software Inc. • www.sisoft.com

6 Clock Tower Place • Suite 250 • Maynard, MA 01754

(978) 461-0449 <978-461-0449> x124  • twesterh@xxxxxxxxxx

*“*I want to live like that”

                                             -Sidewalk Prophets

*From:*ibis-macro-bounce@xxxxxxxxxxxxx <mailto:ibis-macro-bounce@xxxxxxxxxxxxx> [mailto:ibis-macro-bounce@xxxxxxxxxxxxx] *On Behalf Of *Muranyi, Arpad
*Sent:* Thursday, November 2, 2017 2:27 PM
*To:* ibis-macro@xxxxxxxxxxxxx <mailto:ibis-macro@xxxxxxxxxxxxx>
*Subject:* [ibis-macro] Re: Question about "impulse response"


I am trying to apply what you are saying to IBIS-AMI.  Let’s use an example in which

the differential Tx buffer has a 400 mV voltage swing on each of its outputs, and as a

result, the difference step waveform at the Rx goes from -400 mV to +400 mV (i.e.

the amplitude of the step waveform is 800 mV).  The simulator will differentiate this

step waveform and pass it into the AMI Tx Init function as the impulse response of

the channel (or convolve with the output of the Tx GetWave function for the time

domain AMI simulation).  Is this the correct interpretation of “impulse response” in

the IBIS-AMI specification?  Or should the 800 mV difference step waveform be

divided by 2 first before differentiating it to become the “impulse response” for the

purposes if AMI simulations?




*From:*ibis-macro-bounce@xxxxxxxxxxxxx <mailto:ibis-macro-bounce@xxxxxxxxxxxxx> [mailto:ibis-macro-bounce@xxxxxxxxxxxxx] *On Behalf Of *Mike Steinberger
*Sent:* Thursday, November 2, 2017 9:02 AM
*To:* ibis-macro@xxxxxxxxxxxxx <mailto:ibis-macro@xxxxxxxxxxxxx>
*Subject:* [ibis-macro] Re: Question about "impulse response"


If you want an NRZ signal with unit amplitude, the symbol voltages need to be -0.5 and +0.5. The symbol voltages can then be multiplied by the signal amplitude in the system being modeled to correctly represent the signal in that system. Choosing -1.0 and +1.0 is simply an error to begin with, unless one then multiplies by half the signal amplitude in the system being modeled.

My 2C.
Mike Steinberger

On 11/02/2017 12:27 AM, Muranyi, Arpad wrote:

    Todd, Walter,

    I understand that the impulse response represent the analog
    channel, including the Tx and Rx buffers.

    I also understand that the voltage swing the Tx can drive into the
    channel is also captured by this

    channel response, and for that reason we are not talking about
    “unit” *** responses.

    Let me try to explain my question in different words:

    First, let’s assume for the sake of this conversation that the
    channel impulse response is generated

    by taking the derivative of the channel step response waveform. 
    With this assumption we are

    shifting the question to what is the definition of the channel’s
    step response.  Let’s assume that

    a step response is going from logic ‘0’ to logic ‘1’.  The IBIS Tx
    buffer has a digital stimulus internal

    to the simulators which does just that.

    Using that stimulus, the non-inverting output of Tx goes ‘0’ to
    ‘1’ and the inverting output goes

    ‘1’ to ‘0’.  However, in IBIS-AMI we are dealing with the
    difference waveform at the Rx input, in

    which case the states would be ‘-1’ and ‘+1’.  (We all know that
    the difference waveform is usually

    twice the magnitude of the two individual output waveforms of a
    differential buffer).  So if the

    states of the difference step waveform obtained at the Rx
    correspond to ‘-1’ and ‘+1’, while the

    “text book” definition of a step response is ‘0’ to ‘1’, than we
    should theoretically divide the channel’s

    difference step response waveform seen at the Rx input pins by 2
    in order to arrive to a “text book”

    channel step response (which corresponds to the states ‘0’ and
    ‘1’) before we differentiate that

    step response to generate the channel’s “impulse response”.

    But I am afraid that this is not what all EDA vendors do when they
    generate the impulse response

    for the AMI Tx Init function.  I suspect (this needs to be
    verified) that most simulators just take the

    channel step response difference waveform at the Rx and take the
    derivative of that to turn it into

    a channel impulse response for the Tx AMI Init function.

    This is why I wanted to find a definition in the IBIS-AMI
    specification for what “impulse response”

    is, but I didn’t find the answer to my question.  And since I
    don’t like customers saying things like

    “but other simulators do this or that”, I would like to make sure
    that we, EDA vendors, are all in

    agreement on this question.

    This might also have an effect on how we deal with single ended
    channel response waveforms.

    Theoretically those should not be divided by 2, but then how would
    the EDA tool know when is

    the waveform single ended or differential in order to decide when
    the channel response should

    be divided by 2 or not...  Or, should this be decided by looking
    at the [Diff Pin] keyword and check

    whether the [Model] is differential or not?




    *From:*Todd Westerhoff [mailto:twesterh@xxxxxxxxxx]
    *Sent:* Thursday, October 26, 2017 9:48 AM
    *To:* wkatz@xxxxxxxxxx <mailto:wkatz@xxxxxxxxxx>; Muranyi, Arpad
    <Arpad_Muranyi@xxxxxxxxxx> <mailto:Arpad_Muranyi@xxxxxxxxxx>;
    ibis-macro@xxxxxxxxxxxxx <mailto:ibis-macro@xxxxxxxxxxxxx>
    *Subject:* RE: [ibis-macro] Re: Question about "impulse response"

    Arpad & Walter,

    I think the thing that we need to be really clear about in
    IBIS-AMI is that the impulse response represents the analog
    channel, **NOT** the passive channel. I’ll admit to only having
    done a quick search, but the only two places where I find that
    point made are at the start of and the start of (both on page 177). Both of these are indirect
    statements; they require that you understand what the analog
    channel is and how it’s different from the passive channel.

    So – to the original question, I would say it’s “under-defined”
    and could use a bit of beefing up.


    *Todd Westerhoff*

    VP, Semiconductor Relations

    Signal Integrity Software Inc. • www.sisoft.com

    6 Clock Tower Place • Suite 250 • Maynard, MA 01754

    (978) 461-0449 <978-461-0449> x124  • twesterh@xxxxxxxxxx

    *“*I want to live like that”

                                                 -Sidewalk Prophets

    [mailto:ibis-macro-bounce@xxxxxxxxxxxxx] *On Behalf Of *Walter Katz
    *Sent:* Wednesday, October 25, 2017 9:51 PM
    *To:* Arpad_Muranyi@xxxxxxxxxx <mailto:Arpad_Muranyi@xxxxxxxxxx>;
    ibis-macro@xxxxxxxxxxxxx <mailto: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?


    Walter Katz

    wkatz@xxxxxxxxxx <mailto:wkatz@xxxxxxxxxx>

    978.461-0449 x 133

    Mobile 303.335-6156

    [mailto:ibis-macro-bounce@xxxxxxxxxxxxx] *On Behalf Of *Muranyi, Arpad
    *Sent:* Wednesday, October 25, 2017 9:01 PM
    *To:* ibis-macro@xxxxxxxxxxxxx <mailto:ibis-macro@xxxxxxxxxxxxx>
    *Subject:* [ibis-macro] Question about "impulse response"

    Hello Everyone,

    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

    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…




Other related posts: