Re: last_crossing()

Geoffrey,
    David is right, last_crossing() need not control the time step, it
simply records the time when crossings occur, refines them using
interpolation, and then reports the time of the last crossing. However,
you make a good point. It does only make sense that it be in an analog
context so that its argument, which could be an expression, gets
evaluated on every time point.

Sorry, I should have thought a bit more about this before I suggested it.

-Ken

Geoffrey.Coram wrote:
> I thought last_crossing() controlled the timestep in the analog solver
> so that the crossing could be resolved accurately.  Thus, it doesn't
> seem to make sense in a digital context.  If the digital event happens
> a long time after the crossing, the simulator may not be able to
> "back up" to make an accurate measurement.
> 
> Can you make the measurement in the analog domain, and "sample" the
> result when you want it in the event-driven context?  I guess this
> would slow the performance if you don't need the accurate
> measurement most of the time.
> 
> -Geoffrey
> 
> 
> 
> Ken Kundert wrote:
>> All,
>>     Just noticed a deficiency in Verilog-AMS. I went to use
>> last_crossing() to make a very accurate time measurement in an always
>> block, and I could not do it. last_crossing() is only allowed in the
>> analog block. It would be really helpful to also have access to it in
>> the event-driven context.
>>
>> Comments?
>>
>> -Ken
>>
> 

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