Re: Verilog-AMS Committee Meeting Minutes - Oct 5 2006

Dave,
    The existing description of the small-signal transfer
characteristics of the slew function seem perfect to me.

Geoffrey,
    Good catch!

-Ken

Dave Miller wrote:
> Hello Geoffrey, comments on your questions below:
> 
> Geoffrey.Coram wrote:
>> Ooh, I'm really sorry I missed this meeting.  I have a
>> couple comments.
>>
>> Dave Miller wrote:
>>  
>>> Section 4.4.10 - The last sentence in the last paragraph should read:
>>> "In AC small-signal analysis, the slew() function has unity transfer
>>> function."
>>>     
>>
>> Is this right?  I certainly believe that the "usual" ac analysis
>> that follows a dc operating point would have unity transfer
>> function for the slew.  But in some cases, one can do an ac
>> analysis on an operating point captured from a transient.
>>
>> Although periodic-ac or p-noise are outside the LRM, I would
>> expect that, for consistency, one would need to have a
>> non-unity transfer function during these analyses when the
>> function is slewing, and this would dictate a non-unity
>> transfer function for an ac analysis of a transient operating point.
>>
>>   
> It was mentioned that the last sentence of this chapter was hard to
> understand:
> "In AC small-signal analyses, the slew() function has unity transfer
> function except when slewing, in which case it has zero transmission
> through the function."
> 
> We then thought that the last part of that sentence was not needed, I
> guess we were looking at it from a usual ac analysis point of view (I
> was at least).
> 
> Can you suggest an alternative wording for this sentence that would
> cover the condition you mention above, or with this other situation in
> mind, is the original wording of the sentence sufficient?
> 
>>> Section 4.5.4.3 - Graham will check to ensure that vector
>>> parameters are allowed as the first argument to noise_table(),
>>> Also make a mention in this section that a vector can be a
>>> concatenation or vector parameter or combination of both,
>>> as long as the resulting "vector" is even in length (must be
>>> pairs).
>>>     
>>
>> The noise_table() description does not specify what happens
>> for frequencies below the smallest value or above the largest
>> value specified in the vector.  This needs to be clarified.
>> Is it an error?  Does the simulator extrapolate - linearly
>> or with a clamp?  Or is the noise power zero?
>>
>>   
> Sorry, I missed this item in Mantis (Mantis id 0001389). If no
> objections, I can update the noise_table() section with proposal you
> have given in the ticket as it matches 3 Verilog-A implementations I
> know of. This will clamp frequencies to the lower / upper bounds
> appropriately and return corresponding power.
>> Also, there was a request, I believe, to have something other
>> than a linear interpolation, specifically logarithmic.
>>   
> I could not find a ticket for this request, maybe never made it to
> Mantis. I guess this would require a third (optional) argument to
> noise_table() specifying which type of interpolation to use similar to
> $table_model() syntax? Do we want to try and add this into 2.3?
> 
> Cheers...
> Dave
>>
>> -Geoffrey
>>
>>   
> 
>