Subject: Re: [vhdl-200x-mp] Analog/Digital conversion
From: Kevin Cameron x3251 (Kevin.Cameron@nsc.com)
Date: Tue May 06 2003 - 10:59:19 PDT
> From: "Paul J. Menchini" <mench@mench.com>
>
> Kevin,
>
> >> Are the facilities of VHDL-AMS insufficient?
>
> > Since VHDL-AMS doesn't provide access to driver waveforms or
> > "conversion modules" I'd say it's insufficient, but I don't think it's
> > particularly difficult to add that functionality.
>
> I was under the impression that VHDL-AMS had the equivalent of
> conversion modules, but it's been several years since I looked at the
> language in detail. Peter Ashenden may wish to comment....
You'ld need to insert it where a conversion function goes.
> > BTW, is my understanding that VHDL does hierarchical signal resolution
> > correct (a signal is resolved locally and the result passed up, and
> > the effective value propagated back from the top)?
>
> Signal resolution occurs at each level in the hierarchy where a signal
> in the net is of a resolved type. Driving values are propagated upwards
> to the root signal. The driving value of the root signal becomes its
> effective value, and that effective value is then propagated down to the
> leaves.
>
> Paul
That's what I thought, it causes problems with AMS and bidirectional flow
when it's done like that. For hardware modelling you really want to do
resolution flat i.e. you consider all drivers at once and ignore the
design hierarchy. In which case you have to do type conversion differently:
each driver is individually converted to the domain/type of highest accuracy
resolution is performed there and then the result converted back to give
effective values, and to do that you need to mark the types as being
"flat resolution" only and you need a type hierarchy to indicate relative
accuracy (it's easy in Verilog-AMS because there are only two types).
Kev.
This archive was generated by hypermail 2b28 : Tue May 06 2003 - 11:00:16 PDT