Subject: RE: [IBIS-Users] pullup/pulldown range LVDS
From: Mike LaBonte (milabont@cisco.com)
Date: Fri Aug 15 2003 - 12:35:59 PDT
MessageLet's clarify which curves we are flattening. Each characteristic
(high state V/I and low state V/I) is split up into 3 separate curves that
must be combined mathematically by the simulator, and the curves can have
overlapping voltage ranges. At each voltage point the simulator has to add
the current of all 3 curves. But at any given voltage, only 1 or 2 of the
curves has data. The simulator has to extrapolate the other curves to see
what their current would be at the given voltage, assuming the slope at the
end continues on. Alternatively it could assume zero current at all points
beyond the ends of the curves. But the right approach depends on what the
model generator had in mind. For example, s2ibis2 generates curves that
*usually* fall to zero current at the ends, and it "believes" that the
current remains at zero beyond the ends. If it produces curves that do not
fall to zero at the ends, the IBIS model does not work right when an
extrapolating simulator adds the curves together. The end flattening method
that Lynne presents is derived from the way of making it so that the same
result is achieved in either type of simulator, if the current falls to zero
at the end.
Given that, the 2 clamp curves should be flattened at the inner ends to zero
current and slope, and the pullup and pulldown curves should have some
non-zero slope as Andy suggests. Otherwise, current would be constant in the
region between the driver voltage swing and VCC, because the clamps
contribute nothing there. That is not realistic, and simulators might not do
the right thing. Continuing the last known slope is preferable, but if it
results in a final current value at -VCC and 2VCC that the parser doesn't
like, it seems like reducing the slope wouldn't hurt. After all, the
simulation is not expected to go there, and in fact the device is probably
not designed to go there. If a V/I point is altered (in a forest :-) and no
one simulates it, was it ever altered?
Mike LaBonte
-----Original Message-----
From: owner-ibis-users@eda.org [mailto:owner-ibis-users@eda.org]On Behalf
Of Ingraham, Andrew
Sent: Friday, August 15, 2003 1:32 PM
To: ibis-users@eda.org
Subject: Re: [IBIS-Users] pullup/pulldown range LVDS
Dr. Lynne Green gave this as an example:
> Example with 12mA end points in I-V data:
> -3.3V -12m | dummy end point
> 0.4V -12m | end point of actual operating region
> | normal data, up to 96 more lines
> 1.6V 12m | end point of actual operating region
> 6.6V 12m | dummy end point
Personally, and maybe this is just me, I hate it when the endpoints do not
continue the slope of the measured (or simulated) data.
In the case above the regions <0.4V and >1.6V have zero slope. Is this
realistic? Probably not. Only part of the reason for the dummy end points
is to satisfy the IBIS Parser. But it is also there to tell the simulator
SOMEthing about how your device really behaves, to get the simulator back on
track when it wanders out into that region.
Personally I would prefer that you extrapolate the measured or simulated
data points when picking the dummy end points. If the measured/simulated
data has a positive slope, pick dummy end points that are on that slope, or
close to it.
Accuracy is less important when you are that far out of the normal
operating region, as is giving the simulator the right ballpark idea. If
not, you might find that the circuit actually converges at 6V and then you'd
be in trouble.
Regards,
Andy
|------------------------------------------------------------------
|For help or to subscribe/unsubscribe, email majordomo@eda.org
|with just the appropriate command message(s) in the body:
|
| help
| subscribe ibis <optional e-mail address, if different>
| subscribe ibis-users <optional e-mail address, if different>
| unsubscribe ibis <optional e-mail address, if different>
| unsubscribe ibis-users <optional e-mail address, if different>
|
|or email a written request to ibis-request@eda.org.
|
|IBIS reflector archives exist under:
|
| http://www.eda.org/pub/ibis/email_archive/ Recent
| http://www.eda.org/pub/ibis/users_archive/ Recent
| http://www.eda.org/pub/ibis/email/ E-mail since 1993
This archive was generated by hypermail 2b28 : Fri Aug 15 2003 - 12:39:49 PDT