add a "check&save" in the ADE "netlist&run" action ?

[fogh]

Hi All,

did anyone already make an ADE flowchart modification or any other kind of
customisation (trigger, redefine a function by a wrapper...) to get the
schematic to be automatically "checked" (or at least connectivity extracted)
when you click the netlist&run in analog artist ?

Same question with extracted views: a customisation (involving only ADE and no
hierarchy editor, if possible) that lets you "push-button" simulate an extracted
view after modifying or deleting devices in it.

 

[fogh]

Svenn Are Bjerkem wrote:

Haven't done this either, but have a question: Would you first create
the config view with the HEDitor, and then use this config for all the
work in ADE?
I prefer to use no config views at all. Only the switchviewlist in ADE

 

[Svenn Are Bjerkem]

fogh wrote:

I prefer to use no config views at all. Only the switchviewlist in ADE

As a designer I would certainly get a problem with the switchlist alone.
Some cells in a library I want to simulate with schematics, some with
av_extracted and some with external models. It is fairly impossible to
maintain a switchlist so that I get the view that I want in my final
netlist as far as I can see. We have an in-house netlister that take
configs as input, and compared to previous experience, I start liking
these configs, except that cadence haven't given us total freedom to
create a netlist across cellviews and libraries.

--
Svenn

 

[Svenn Are Bjerkem]

fogh wrote:

did anyone already make an ADE flowchart modification or any other kind
of customisation (trigger, redefine a function by a wrapper...) to get
the schematic to be automatically "checked" (or at least connectivity
extracted) when you click the netlist&run in analog artist ?

No, but I would be interested in one. I have managed to bind the netlist
& run to my . key with sevNetlistAndRun(sevSession(hiGetCurrentWindow)).
It would be no problem to add the schHiCheckandSave() in a wrapper and
you actually gave me an idea there...

Same question with extracted views: a customisation (involving only ADE
and no hierarchy editor, if possible) that lets you "push-button"
simulate an extracted view after modifying or deleting devices in it.

Haven't done this either, but have a question: Would you first create
the config view with the HEDitor, and then use this config for all the
work in ADE?

--
Svenn

 

[fogh]

Svenn Are Bjerkem wrote:

fogh wrote:

I prefer to use no config views at all. Only the switchviewlist in ADE
As a designer I would certainly get a problem with the switchlist alone.
Some cells in a library I want to simulate with schematics, some with
av_extracted and some with external models. It is fairly impossible to
maintain a switchlist so that I get the view that I want in my final
netlist as far as I can see. We have an in-house netlister that take
configs as input, and compared to previous experience, I start liking
these configs, except that cadence haven't given us total freedom to
create a netlist across cellviews and libraries.

OK I see. It was simply my opinion, probably because I did not see much demand
here for these heavily "mixed-level" simulations. So we cope with the library
manager. For instance, if you want only 1 block to have parasitics, rename
schematic to schematic_actual, and copy av_extracted to schematic. Idem if you
want to simulate all behavioural except 1 cell transistor level.

I like to keep design flows straightforward: a step that is useful 1% of the
time I do not like to introduce as systematic. When people get used to just see
boxes pop-up and click them away without knowing what was written there, it is
IMNSHO the beginning of the end, in terms of risks.

 

[Svenn Are Bjerkem]

fogh wrote:

OK I see. It was simply my opinion, probably because I did not see much
demand here for these heavily "mixed-level" simulations. So we cope with
the library manager. For instance, if you want only 1 block to have
parasitics, rename schematic to schematic_actual, and copy av_extracted
to schematic. Idem if you want to simulate all behavioural except 1 cell
transistor level.

I have worked in different environments with cadence. The most difficult
thing is to remember what I did with the design 2 years ago, or even
worse, what did the colleague do 2 years ago and he is no longer here.

Wenn you use a revision control system in the background, it is not very
much wanted that views get copied around. Needless to say, such systems
are used in environments where many engineers work on and with the same
design. And in a global player, face to face communication is not always
possible.

I like to keep design flows straightforward: a step that is useful 1%
of the time I do not like to introduce as systematic. When people get
used to just see boxes pop-up and click them away without knowing what
was written there, it is IMNSHO the beginning of the end, in terms of
risks.

I hate modal dialog boxes that says something is ok, please press ok.
That is what a console window like CIW is there for. I like the command
line. I love make because I can make a procedure reproducable. If it
worked yesterday, it has to work today unless somebody has changed
something. If somebody has changed something, I want to know _who_ and
_when_, if possible also _why_.

A config view is something close to make. I can create as many config
views I like in my sandbox environment and be able to create a
simulation environment that contains exactly what I want without needing
to change any rules that I forget to change back next time. The config
view work as a documentation which modules were used during a particular
simulation. This is nice to have when somebody wants to know what went
wrong. (Nobody want to know why something went well)

Cadence could extend the HEDitor to allow me to combine *any* view of
*any* cell from *any* library independent of what the schematic
hierarchy instantiations say. One could introduce levels of relaxations
in order to help the designer towards a consistent library tree so that
a tape-out simulation would only be accepted when all view are the same
and come from the libraries actually instantiated. In a large company
there are so many different opinions on what is a good design flow so
somebody has to cut through and define something. The config view gives
a bit of freedom back to the designer. That is at least my experience
the last few months.

--
Svenn