Walter, Regarding "AMI simulations assume that there is a zero impedance voltage source at the Tx end of the analog model and an infinite impedance Probe at the Rx end of the analog model.", unfortunately this assumption is not spelled out in the specification. Considering that "According to the reference flows, to do a time domain simulation, the waveform that is input to Rx AMI_GetWave is the waveform generated by Tx AMI_GetWave convolved with the impulse response of the channel" the implication is that the GetWave flow deals with signal processing in which there are no voltages and currents or Ohm's and Kirchhoff's laws, etc..., i.e. there is no circuit simulation, only signal number crunching. We think of these signals as a voltage vs. time waveform, but in reality they could be anything, such as the CO2 content of the atmosphere vs. the number of cars sold. As long as we don't spell out the assumption you stated, we can only talk about signal processing in the AMI_GetWave flow which excludes the alternate way of doing the AMI_GetWave flow as you described below. Regarding "This method is not available with the IBIS analog model (either Legacy IV, or External Model) because of the interpretation of IBIS models in External Model that they have digital inputs", we can't blame the [External Model] for this. I am preparing a presentation of the next ATM meeting and I will try to show a few interesting details on this topic. Thanks, Arpad ==================================================================== From: ibis-macro-bounce@xxxxxxxxxxxxx [mailto:ibis-macro-bounce@xxxxxxxxxxxxx] On Behalf Of Walter Katz Sent: Friday, December 21, 2012 8:38 AM To: IBIS-ATM Subject: [ibis-macro] AMI Analog Modeling All, AMI modeling assumes there is an analog channel that can be represented by an impulse response. AMI simulations assume that there is a zero impedance voltage source at the Tx end of the analog model and a infinite impedance Probe at the Rx end of the analog model. The impulse response of the channel can be generated a number of ways. One way is to do a time domain step response of the channel, the impulse response is the derivative of the waveform at the Rx. A second way is to cascade into a single s4p the Tx buffer analog model represented as an s4p, the channel (die to die) represented as an s4p and the Rx buffer analog model represented as an s4p. The two impulse responses should be identical (within numerical precision/accuracy). There can be other methods of generation this impulse response. According to the reference flows, to do a time domain simulation, the waveform that is input to Rx AMI_GetWave is the waveform generated by Tx AMI_GetWave convolved with the impulse response of the channel. An alternative method of generating the waveform at the Rx AMI_GetWave is to do a time domain simulation using the waveform generated by Tx AMI_GetWave (mad differential) as input to the cascaded s4p defined above, and probing the waveform at the Rx end of the cascaded s4p. This method is not available with the IBIS analog model (either Legacy IV, or External Model) because of the interpretation of IBIS models in External Model that they have digital inputs. One really needs to understand these statements to understand how AMI analog modeling is fundamentally different than the existing interpretation that IBIS models have digital inputs. Walter Walter Katz wkatz@xxxxxxxxxx<mailto:wkatz@xxxxxxxxxx> Phone 303.449-2308 Mobile 303.335-6156