Re: Minutes: Ex1

[Sorry I missed the call]

Folks who are not familiar with how analog/ms simulators work should
note that the it is actually sampled rather than continuous. The
simulator will attempt to evaluate future state as far ahead as possible
for a given start state, and will step through conditions unless they
are specifically mentioned in "cross" statements. This is mostly
required for efficiency since automatically stepping to exactly where
(say) "V(a) < 10.0 mV" changes may cause unnecessarily short timesteps.

Signals in Verilog-AMS are both analog and digital: at any given time a
wire has both an analog value and a digital value. The automatic
insertion of A to D converters in Verilog-AMS means that in any context
the digital value of a signal "x" and its analog representation "V(x)"
should be in sync (if the conversion code is written correctly). This is
a feature of the language which allows "plug & play" of analog and
digital modules without incurring timing errors.

@cross is both digital and analog in that it is evaluated in analog to
stop at the right time but you can perform digital functions in the
@cross block since it happens at a discrete time.

Kev.

Anand Himyanshu wrote:
> ....
>  
> Discussion on examples followed.
>  
> Ex 1:
> Ken asked the question whether V(a) < 10.0 mV goes from false to true
> in continuous time or digital time? There was some discussion on this.
> Himyanshu mentioned that in the example, the transition was envisioned
> with ideal semantics in continuous time. Ken mentioned that the value
> of the expression is Boolean. Value of the expression at 'Time Zero'
> was also discussed. Dejan mentioned that the property could be
> rewritten using past time. John asked there were questions about the
> interpretation of past time at Time Zero. It was felt that 'past or
> negative time' needs to be discussed more, but for this example we had
> thought about future time. Since there was no governing event and in
> the example we avoided the use of ##0 and instead used #0.
>  
> Is 'c' signal digital or analog signal? Himyanshu mentioned it could
> be viewed as a digital signal. John mentioned it could be either
> and no claims were being made whether it was digital or analog, as
> long as the values are available in the continuous time domain. Ken
> mentioned that if 'c' is digital then at every increment of the
> digital simulator its value is defined. It needs to be mapped
> (aliased?) to analog domain or vice versa. There was a good deal of
> discussion on digital and analog values and how they interacted with
> each other within an AMS simulator.
>  
> Ken asked another question, Is @cross, digital or analog? John
> mentioned we had not thought about it and assumed ideal semantics. It
> was mentioned that @cross behaved differently depending upon whether
> it was digital or analog and that needs to be taken into account.
>  
>


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