C
Chaos Master
- Jan 1, 1970
- 0
Anybody here knows any program to convert a netlist from SPICE, to a schematic
(i.e. the reverse of a netlister)?
(i.e. the reverse of a netlister)?
Chaos said:Anybody here knows any program to convert a netlist from SPICE, to a
schematic (i.e. the reverse of a netlister)?
Amazingly difficult task. Have a think just what is involved tpo to
this?
Amazingly difficult task. Have a think just what is involved tpo to
this?
However, http://www.concept.de/ has one. Apparently.
Paul said:I find I can do this relatively easily on paper for individual stages,
but it does get increasingly difficult the more components/stages
there are.
Kevin Aylward said:Amazingly difficult task. Have a think just what is involved tpo to
this?
However, http://www.concept.de/ has one. Apparently.
Terry said:I've never really understood why it should be so difficult. Every
part, every node and pin is identified unambiguously. The two
relatively difficult aspects presumably are:
1. Drawing the connections in an 'inteligent' fashion, minimising
cross-overs, etc.
Kevin Aylward said:Ahmm...That's because you aint really thought about it much.
Try actually having a go of outlining the software to do this. Say
you've got 1000 transistors in a text file. How do you go about
"knowing" a counter from a monostable, or from a differential amp. The
combinations are limitless. You need to generate significant pattern
recognition.
Terry said:But you have?
Why would such a program need to tackle any of the irrelevant issues
you've raised?
At the level I described, why does it need to recognise
and label the configuration of say a 555 plus a few Rs and Cs as 'a
monostable'?
My assumption is that the user, studying the output,
might recognise it as such - but even that's not essential to build
it.
Presumably your next objection in support of your original
superficially
-considered objection to the OP will be: 'Ahmm - Just
think about it. How could a program tell what size the components are
and where to bend the wires before soldering them?'
[snip the ad that Kevin should, by convention, dash-dash-space][snip]Terry said:But you have?
Kevin Aylward
Jim said:[snip the ad that Kevin should, by convention, dash-dash-space][snip]Terry said:Ahmm...That's because you aint really thought about it much.
But you have?
1. Drawing the connections in an 'inteligent' fashion, minimising
cross-overs, etc.
Try actually having a go of outlining the software to do this. Say
you've got 1000 transistors in a text file. How do you go about
"knowing" a counter from a monostable, or from a differential amp.
The combinations are limitless. You need to generate significant
pattern recognition.
Kevin Aylward
I have traced a few simple-minded IC's by "drawing" the (numbered)
devices into an array, the connecting them as I traced thru the metal.
I then used "rubber-banding" movements to bring some order to the
schematic.
All-in-all an enormous MANUAL task, typically taking a week...
financially viable because it was a patent infringement issue and I
was being paid "lawya" rates.
I think automating it would be worthy of a Nobel prize ;-)
Kevin Aylward said:You don't know what your talking about. Non of what I said is
irrelevant. What exactly do you think is irrelevant? To make sense a
circuit diagram has to be in a reasonable conventional format. Do you
want me to send you a 200 transistor netlist so that you can actually
try it in real life how hard it is to make a readable schematic? How do
you knoew what block is associated with what block. e.g. suppose you
have a PLL, consisting of a 100 transistor digital phase/frequency
detector, a 50 transistor VCO, a 50 transistor charge pump etc...How do
you think a program is going to automatically assemble all of this into
standard cong=figurations.
You out to lunch. Sure, an ittsy bittsy 555 timer circuit may well be
simple as a one off. You try it with a circuits that has any sort of
complexity. Its a huge task to go from a rats nest to a readable
schematic.
Sorry, mate, you don't seem to have a clue. This is a really difficult
problem. If it were that easy, there would be lots of them out there.
Wake up dude. Go and actually try it before you make such daft cliams.
Terry said:OK, it's plain that we're respectively talking about circuits at
different ends of the complexity spectrum. Maybe ChaosMaster is
working with designs on the scale you describe. I assumed we were
closer to *my* end of the scale.
FWIW I *have* tried it. Manually it's a chore, but I made useful
inroads. I expect the data processing necessary would increase at a
combinatorial rate. But nevertheless, given today's CPU speeds, IMO a
reasonably efficient program could break the back of what *I* would
regard as a fair size circuit.
Jim said:[snip the ad that Kevin should, by convention, dash-dash-space]Terry Pinnell wrote:
[snip]Ahmm...That's because you aint really thought about it much.
But you have?
1. Drawing the connections in an 'inteligent' fashion, minimising
cross-overs, etc.
Try actually having a go of outlining the software to do this. Say
you've got 1000 transistors in a text file. How do you go about
"knowing" a counter from a monostable, or from a differential amp.
The combinations are limitless. You need to generate significant
pattern recognition.
Kevin Aylward
I have traced a few simple-minded IC's by "drawing" the (numbered)
devices into an array, the connecting them as I traced thru the metal.
I then used "rubber-banding" movements to bring some order to the
schematic.
All-in-all an enormous MANUAL task, typically taking a week...
financially viable because it was a patent infringement issue and I
was being paid "lawya" rates.
I think automating it would be worthy of a Nobel prize ;-)
...Jim Thompson
Amazingly difficult task. Have a think just what is involved tpo to
this?
However, http://www.concept.de/ has one. Apparently.
sorry, kevin. that reply was for terry. i waited for it to show up so iAmazingly difficult task. Have a think just what is involved tpo to
this?
However, http://www.concept.de/ has one. Apparently.
Kevin Aylward
[email protected]
http://www.anasoft.co.uk
SuperSpice, a very affordable Mixed-Mode
Windows Simulator with Schematic Capture,
Waveform Display, FFT's and Filter Design.
I've never really understood why it should be so difficult. Every
part, every node and pin is identified unambiguously. The two
relatively difficult aspects presumably are: