FreeLists / si-list / [SI-LIST] Re: I-V vs V-T endpoints In IBIS

["Ingraham, Andrew" <Andrew.Ingraham@xxxxxx>]

>  I know that a load line is 
> applied to the I-V data and the enpoints of this are matched to the 
> endpoints
> of the V-T data.  I'm not sure of the importance of why these
endpoints 
> should match.
 
Actually, the intersections between the load line and the I-V data (not
the endpoints!) are checked against the endpoints in the V-T data.

The I-V data + load line intersections, should give you the initial
voltage (first V-T datapoint) and the final voltage (last V-T
datapoint).  If they don't, it generally means something is wrong.  The
first V-T datapoint really should match, with few exceptions.  The last
V-T datapoint should come close, as long as the V-T table extends out
far enough in time to reach the "end" of the transient.  If it hasn't
died out by the end of the table, then the data was truncated.


>  My understanding is that the I-V data is used to model the 
> endpoints 
> of the device, where the V-T data is used to model the low-to-high and

> high-to-low transitions. I know that in HSPICE, when there is a
mismatch, 
> the V-T 
> endpoint is modified to match the I-V endpoint. So how does this
warning 
> affect the accuracy of the model? I'm thinking the V-T data plays a
bigger 
> role than I think? 
 
What a simulator does with the V-T data, with or without the warning, is
very implementation dependent.

One simulator might issue this warning but use the data anyway.  Another
might adjust the data, as you say HSPICE does.  Either way, the results
may or may not have any relationship to reality (or to a SPICE model
from which this IBIS data was generated).


>   And you can use the Ramp data and get the same result without 
> even using
> the V-T data.
 
That is probably very device-dependent, as well as simulator-dependent.

Some devices might come close to the "default" behavior from a given
simulator when there is no V-T table, but another might not.  That's why
we have V-T data, so that you can tell the simulator exactly how a
particular component behaves when switching.

Without V-T data, the simulator might do something like a linear ramp
from pull-down I-V curve to pull-up I-V curve, turning off one curve
while simultaneously turning on the other.  Maybe that's close enough,
and maybe it worked well with the devices you've tried.  But it is
probably not a good match for other devices.

Say you have a device where one FET turns off a few nanoseconds before
the other turns on.  Or another where both FETs overlap briefly (lots of
"crowbar" current).  Using only the I-V curves, the two devices could
yield identical simulations, yet behave very differently.  A device with
significant capacitance from the FET gates to the outputs, causing a
feed-forward glitch in the opposite direction, would not show this
behavior without V-T curves.


>   I have a customer who is asking about this, and I want to 
> tell them that it's okay to have the warnings because of the above 
> reasons, but I'm 
> not entirely sure about that. 
 
The warnings usually mean something is wrong with the model.  Maybe a
judgment call is appropriate if the error is borderline, but otherwise
I'd try to get the model fixed before putting any trust in it.

Regards,
Andy



------------------------------------------------------------------
To unsubscribe from si-list:
si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field

or to administer your membership from a web page, go to:
http://www.freelists.org/webpage/si-list

For help:
si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field

List archives are viewable at:     
                http://www.freelists.org/archives/si-list
or at our remote archives:
                http://groups.yahoo.com/group/si-list/messages 
Old (prior to June 6, 2001) list archives are viewable at:
                http://www.qsl.net/wb6tpu