[Author Prev][Author Next][Thread Prev][Thread Next][Author Index][Thread Index]
Re: gEDA-user: problem
On Sunday 05 March 2006 15:01, Stuart Brorson wrote:
> For PCB layout, there is no tool, at least as far as I know.
> The reason is that the PCB fab process is less well
> characterized & more variable than the IC process. Also,
> real models for the parts used don't exist. Most op-amps,
> etc., provide only macromodels, which don't capture this kind
> of info.
There are several commercial tools. I don't know of any free
ones. The commercial tools are quite sophisticated. They do
waveforms, essentially one trace at a time, with the neighbors
are "aggressors". They model waveforms, crosstalk, and
radiation.
They are basically circuit simulators, not necessarily Spice
based. The best ones are not Spice based. They model
transmission lines very accurately, but are weak on other
devices. Mostly they use "IBIS" models for the drivers and
receivers. Most devices do have appropriate models published.
IBIS models define digital devices as analog behavioral models,
using a pre-defined topology. The model consists mainly of
tables, showing things like waveforms and I/V characteristics.
The tools for generating the models, based on simulations or
measured data, are free/open-source. Most of the models are
generating using one particular tool, that generates a first
cut model. They usually require some manual tuning, but many
modelers don't bother, which leads to some strange models.
Supporting IBIS in a simulator is much harder than would be
expected. One issue is that the waveforms are specified with
real loads, like 50 ohms, 24 pf, to + 5 volts. Usually two
rising and two falling waveforms are supplied. The simulator
needs to figure out how to do it. One commercial one that I
know of (very well) does a multi-dimensional fit and will
handle any number of tables with arbitrary loads.
IBIS also models the passive parts of a device, including the
package and connector. There is also a standard for connector
models.