simon2
Junior Member
Offline
Posts: 27
Southampton, United Kingdom.
|
Hi thlcak, like you I have used Hspice a long time - so long in fact I forgot how whilst using spectre for 12 years, then I had a chance to return to Hspice after becoming Spectre-centric via advice, celerity, eldo, pspice then more recently winspice.
For what it is worth, in terms of producing reliable, believable results most easily I would rate the tools thus: Hspice, Eldo, WinSPICE, Pspice, then trailing by a long way, Spectre. (Sorry Andrew, I know you are a great proponent of Spectre ...).
The reason I say this is that it seems that spectre is good at what it does, which is to provide an analogue interface into the Cadence tool suite and as such is intended (and may be the only) tool which can be used across multi-national, multi-site, multi-discipline teams. And it does that well.
What spectre does not well do is address what we analogue designers really need: to be able to directly and simply (and I stress simply as it often takes two days to work out how to do in spectre-cadence what can be done in hspice or WinSPICE in two or three minutes) and that is to access all of a device instance's parameters and to be able to manipulate them. I won't comment on how poor the scripting capabilities of spectre really are as I am likely to get sued!
Speaking from many hundreds of hours spent sorting out the issues for all of the above tools, I would say that on balance, stick with hspice, its results are far more accurate, despite supposedly using the same model files!
Remember Hspice is still the "gold standard" for the foundries, who at the end of the day underwrite the accuracy of the designs we do.
Perhaps my last comment could be that as a small company on a limited budget where these tools directly effect our "bottom-line" we have chosen to invested in Hspice, not Spectre.
- Hope my opinions haven't offended anyone; they are personal, but have been formed from years of practical (and pragmatic) experience of using the tools "in anger".
Cheers, SimonH.
|