Hi Arpad, Thanks for the reply. Please see if understand the flow with AMI. A .DLL and a .AMI represent the RX and TX algorithmic model. They are "hooked" up with a TX IBIS model that includes a TX package model, also a EBD or other model that represents the PCB, and then to a RX IBIS model which includes a RX package model. A pulse is given to get the channel response and then the AMI simulations can begin. Is this correct? As I understand it an IC vendor would develop the .DLL and .AMI file along with the IBIS file which would include a package model. This model "bundle" (.ibs, .ami. .dll) would be supplied by IC vendor to OEMs for channel analysis. The package model currently is only coupled/uncoupled RLC data. If IC vendor 1 is a supplier of a TX device and IC vendor 2 is a supplier of a RX device and a OEM user wants to model the channel with IC vendor 1 and IC vendor 2 parts, are they relying on the IBIS package model for that portion of the channel? Doesn't the IBIS model make calls to the AMI DLL for the AMI simulation? When it does is it using the package RLC data as part of the analog portion of the channel response or is it looking elsewhere for a S parameter model? Mike DeVita Staff Engineer, Application Engineering & Characterization VM 480 821-2408 ext 5021 FAX 480 821-1713 EM mike.devita@xxxxxxxxx Sent by: <ibis-macro-bounce@xxxxxxxxxxxxx> "Muranyi, Arpad" <Arpad_Muranyi@xxxxxxxxxx> 11/11/2009 10:53 AM Please respond to Arpad_Muranyi@xxxxxxxxxx To "IBIS-ATM" <ibis-macro@xxxxxxxxxxxxx> cc Subject [ibis-macro] Re: Question on package models Mike, AMI is kind of unrelated to the package model. AMI is an algorithmic model which uses a channel response and processes it with statistical algorithms. The reason I said "kind of" is because for generating a channel response, you have to have a full description of the channel, including the Tx, Tx_pkg, T-lines (connectors vias, etc), Rx_pkg, Rx. This is what we call the analog simulation which you need to do once to generate a channel response, but this is usually an independent simulation step which precedes the AMI "simulations". Arpad ============================================================ From: ibis-macro-bounce@xxxxxxxxxxxxx [mailto:ibis-macro-bounce@xxxxxxxxxxxxx] On Behalf Of Mike DeVita Sent: Wednesday, November 11, 2009 11:00 AM To: IBIS-ATM; ibis-macro-bounce@xxxxxxxxxxxxx Subject: [ibis-macro] Question on package models Hi, How does the IBIS package model (either embedded or stand alone) work with a AMI based link? Is it still expected to provide RLC data coupled and/or uncoupled? I do not think this is sufficient to correctly model the package portion of the channel for use with an AMI block/high speed SERDES. Mike DeVita Staff Engineer, Application Engineering & Characterization VM 480 821-2408 ext 5021 FAX 480 821-1713 EM mike.devita@xxxxxxxxx