Re: Regarding ddx operator location in grammar

The Hicum bipolar transistor model uses ddx() in the expression
for the avalanche current. Iavl depends on cjci, which is
the derivative of qjci, the internal base-collector charge.
Thus, gmavl requires the simulator to compute a second
derivative.

The second derivative is no harder to compute than the first;
I think the simulators just need to be fixed.

-Geoffrey

Xavier Bestel wrote:
> On Wed, 2010-10-06 at 08:43 -0500, David Miller wrote:
>> This was what I was suggesting for the ddx() operator.
>>
>> Currently it is lumped with the analog_filter_function_call which contains
>> transition/ddt/idt/slew etc.
>> The analog filter functions have a lot of restrictions, so in the document we
>> have all these cases where we apply these restrictions *except* if the filter
>> function is ddx.
>>
>> I suggest that we pull ddx() out to its own syntax item:
>>
>> analog_partial_derivative_function_call ::=
>> ddx ( analog_expression , branch_probe_function_call )
>>
>> and make analog_partial_derivative_function_call part of analog_primary.
>
> Do we want to emphasize the fact that ddx() should be reserved to
> expressions not involved into contributions ? I have a feeling that
> second derivatives (which are needed only in the ddx() case) won't work
> that well with all simulators. The LRM already notes that internal
> unknowns created by the simulator may also break ddx().
>
> Xav
>
>

-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.