RE: about IO_Buffer model's ac analysis

Sorry, this was sent prematurely when a keyboard key got stuck.

You wrote:

> I think that ibs model was generated by spice's dc analysis and transient
> analysis,
> so ibs model don't contain frequency response information.
 
> Frequency response information IS available in an IBIS model for its
> package parasitics, and I think that's what matters here.
>
> When I do an AC analysis, I am interested in the transfer function FROM
> some point TO some other point in the circuit.
>
> The first thing you need to think about, is where might the FROM and TO
> nodes be when you have IBIS models?
>
> It makes sense to look at transfer functions that might involve the
> device's pad node, either FROM (driving out), or TO (as an input). The
> IBIS model ought to have sufficient information about the device's package
> parasitics (between the pad and the pin), for the simulator to be able to
> do an AC analysis, within the accuracy limits of that package
> representation.
>
> For a buffer driving out, you would replace the I/V curves with your AC
> source. For a receiver, find the slopes of the I/V curves where they
> intercept the DC operating point, and include that resistance, along with
> c_comp, as the "load" on the end of the package.
>
> It doesn't make sense to try to analyze THROUGH a buffer that is
> represented by an IBIS model. The IBIS model of a buffer has no input, so
> there is no such thing as a transfer function that includes that of the
> buffer itself. (HSPICE's B-elements do have inputs, but they are just
> there to make the B-elements work.)
>
Regards,
Andy

 
Received on Tue Mar 27 09:23:09 2001