|
|
|
|
Design package Beauty Contest? - RF Cafe Forums
|
Graham
Post subject: Design package Beauty Contest? Posted: Tue
Oct 11, 2005 7:40 pm
Colonel
Joined: Fri Sep 02,
2005 7:25 pm
Posts: 34
Location: Hampshire UK
Forgive me guys
if this has been well thrashed out somehere already, but.. I couldn't
find it. Before I invest again, and regret, I would like to have some
kind of RF-Cafe beauty contest for EDA design apps.
Selfish that
I am, I hope to duck spending any more of my life going up yet another
lock-you-in learning curve with demos, I would like to tap in to what
some of you have already know about suitability. So, here is some of
what I find in my trawl. I confess some of my judgements here are inexperienced,
and may not be fair, but you can tell me when I am wrong about a product
eg. Microwave Office - some kind comments about it, but is it primarily
aimed at MMIC designers, or would it be just right for someone who wants
to figure a EPHEMT design at 2.4GHz with feedback? Does one need to
have Analog designer as well, to figure the bias, or can it be used
in a more general sense. Should I try the demo without knowing the price?
APLAC, is very capable, but student version hits a memory restriction
on a one transistor circuit with 2 resistors for a bias chain, one in
the collector, a current measurer in the collector, and one swept AC
input, plus a voltage source to run on. Sweep happens OK, but it won't
play if you want the simple DC condition as well. They do something
special about the maths. APLAC comes in lots of separated bits (er..
modules). Its very fast! Cost of ownership..? I am scared to ask. Its
a bit geek to use, and the manual pdf is dripping with boardroom tech
thesis newspeak that could have been honed at a university.
MultiSim,
ex Electronics Workbench, which I have used for a project, is quite
friendly. Nerds would feel patronised (the waveform display does not
have to be on a scope graphic with front panel controls!) BUT.. its
a SPICE simulator unable to do s-parameters. I think it is among the
most popular, but I am not sure of its usefulness for RF.
SWCAD
III is a SPICE simulator free from Linear Technology, with no memory
or other restrictions. It is blindingly fast, and very geared to accurately
modelling waveforms from switchmode power supply designs. Linear provide
lots of them. BUT.. the things one has to do to get a external model
into it, is kinda awkward. Still, it is relentlessly Berkley Spice,
and so has possibilities, provided one can persuade the model parameters
out of suppliers like Agilent.
Ahh.. Agilent. That means ADS
I think. So how good is it? What does it cost? Ballpark will do - I
don't want to tell my life story and describe what technology areas
I might be designing in, just so I can "register" with my "area manager".
From Agilent, it is unlikely a small outfit like me could afford it,
but I will look at anything. I notice that the model for an Agilent
PHEMT discrete comes as a ".zap" file. Renamed to ".zip", WinRar extracts
it to reveal innards which include what looks like SPICE text FET parameters
wrapped up in there. The GUI display files look a bit ORCAD, but I don't
know.
Protel - taken over or become Altium. I knew it as Protel99SE
, but I don't work for the firm that used it anymore. Maybe they fixed
the strange way the schematic could only be navigated using scrollbars,
while the PCB layout moved by mouse "grab 'n shove". It made some complicated
PCBs for me, but the only way to get the autorouter to be useful was
to spend more time setting up its rules than you might have spent using
good old human sense. RF and autorouters are a subject by itself.
Ultiboard? er.. I don't know them
OrCAD, certainly famous.
Are they Mentor graphics/ I didn't keep up with it all, and I couldn't
afford it anyway.
So.. first we need a LIST. Then a formula for
goodness factor. eg. we divide the rating by the price in dollars. We
can multiply by a positive loading if it does RF goalseeking automatch,
etc We can maybe come up with something the industry will come to respect
and fear. The RF-Cafe rating! Hmm - maybe Kurt would not wish it to
be too intimidating.
Top
Kirt Blattenberger
Post
subject: Posted: Wed Oct 12, 2005 10:24 am
Site Admin
Joined: Sun Aug 03, 2003 2:02 pm
Posts: 308
Location:
Erie, PA
Greetings Grahm:
If you would create a poll question
and list of answers choices, I will be glad to use it for the next new
poll.
It would be really useful to have a public forum for constructive
criticism and/or praise for the very expensive design software available
these days.
_________________
- Kirt Blattenberger
RF
Cafe Progenitor & Webmaster
Top
Kirt Blattenberger
Post subject: Registered Users Can Now Create PollsPosted: Wed Oct
12, 2005 12:08 pm
Site Admin
Joined: Sun Aug 03,
2003 2:02 pm
Posts: 308
Location: Erie, PA
Greetings Grahm:
In response to your post, I have modified priviliges for REGISTERED
users like yourself to allow creation of polls in all but the Company
Ratings, Polls, and Webmaster forums.
Please try your hand at
creating a poll for your CAE package topic and let's see how it works.
Thanks.
_________________
- Kirt Blattenberger
RF Cafe
Progenitor & Webmaster
Top
IR
Post subject:
Posted: Wed Oct 12, 2005 4:08 pm
Site Admin
Joined:
Mon Jun 27, 2005 2:02 pm
Posts: 373
Location: Germany
Hello
Graham,
Each EDA tool has its own advantages and disadvantages.
It all depdns on what your field of design is.
I have been using
in many of the tools you mentioned. In the end-effect you will choose
the right tool for your needs by how much money you can spend, what
are your applications/ products etc. The trend nowadays, (And I am sure
that many will agree with me), is that EDA vendors make their tools
integrated so they can support in many applications as possible. This
is exemplified with RF tools as MWO or Eagleware (now Agilent ) that
can do: EM, linear, HB, layout and system simulation all together in
one environment. This is a great advantage intended for users that can
make a flexible choice tailored for their specific needs out of the
full EDA suite.
In summary, you will need to do a serious market
survey and define exactly what are your needs and how much you can afford
to spend in order to find the best choice for your applications/products.
_________________
Best regards,
- IR
Top
Graham
Post subject: Posted: Thu Oct 13, 2005 6:12 am
Colonel
Joined: Fri Sep 02, 2005 7:25 pm
Posts:
34
Location: Hampshire UK
Hi Kurt & IR
It is now in
thought. The point about each package having its strengths in different
areas of application is taken. I should like to contrive a way this
can be reflected so that folk perusing collected information can see
what suits their need. A trite league tabel approach would produce misleading
impressions. (eg the use of league tables in UK school exam results
does not deal well with say a group of 2000 schools, all of which score
beyond 95%. One of them is at the bottom, despite being excellent!)
I am alarmed at the way the semiconductor supply companies try to
aquire, then vary the usefulness of a tool for any but their own product.
I cite as example the original SPICE engine development. This was publicly
funded, and hence was public domain. Understandably, many enterprises
produced front ends and back ends and bells and whistles. Some re-wrote
the code to include traps for convergence problems, and sparse matrix
solution to make the thing faster. If ever there was a need to keep
a dynamically developing, but agreed standard for device model descriptoin
across the industry, it was here. Some offshoot formats became *so*
popular that they became "industry standard" by default (eg. Touchstone).
Looking at an APLAC file, it is clear those Finnish academics tried
hard to make a format that maintained the compatibility with the large
base of SPICE model files out there. They used the same rules, while
introducing their own additional commands and analysis devices.
OK then, before we start "rating" them, I will start by collecting
a list of the packages I think significant enough to include, and I
will respond amicably to anyone making a good case to include any that
I left out in ignorance.
Kurt - somehow I get the feeling I don't
know where this is going to end up!
Graham
Top
Graham
Post subject: Posted: Sun Oct 16, 2005 8:04 am
Colonel
Joined: Fri Sep 02, 2005 7:25 pm
Posts: 34
Location: Hampshire UK
I have been trawling through the sites, and
so far my impression is that more and more of the tools, already hard
to afford, are being mopped up in "consolidations". The tools are either
completely supplied by the device vendors, or gathered into EDA design
house companies with a tight relationship to them.
The new relationships
deriving from recent takeovers is confusing.
ADS is Agilent's
product
EAGLEWARE-ELANIX are still there, now part of Agilents product.
I now have to imagine them competing madly to sell to potential ADS
customers!
ALPAC, now owned by AWR, who are surely going to let
us have the chioce of MICROWAVE OFFICE, with both products steaming
along at full strength - er..are they? Maybe "the new product combines
the strengths of of ..blah ..SYNERGY..blah." OK - I got cynical just
then. Onward..
These are just preamble obsevations as I try to collect
the list together. We will not be comparing like with like, and I would
like it that we end up with something more than a price list. So I put
aside for the moment the big fellas, and take a look at a giveaway.
Linear Technology's SWCAD is a industrial strength SPICE simulator
given away free. The thing has been tweaked to be an extremely fast
product demonstrator, that comes with a whole heap of ready-made applications
using Linear's switchmode PSU devices. The tool allows designers to
alter and adapt the designs to their particular need, (though I suspect
for some designers, the difference is tokenly marginal). Its more than
a tad awkward to generate and apply models for other vendor's devices
in SWCAD, so presumably its value as a design tool for the competition
is limited. SWCAD can be used as a general purpose SPICE simulator,
even for microwave, but handicapped by the model application limits.
You have to be prepared to get dirty with model text files if you wand
it useful.
Here is where we come to the industry leverage. SPICE
as a network calculation method is great, but it has its pitfalls. It
will have a transistor deliver megawatts without complaint if all we
started with was a linear model derived from s-parameters. It will not
calculate past a capacitor-isolated node in a matching network unless
you give it a giga-ohm path to ground to avoid a divide-by-zero. Tools
that have "nice" front-ends to get past these problems, come also with
non-standard formats.
I am still gathering the list. If anyone
can shorten the task of discovering the cost of ownership of these packages,
do tell. When I ask, vendors want my life story + description of intended
projects before the will even send me invites to "seminars"
Posted 11/12/2012
|
|
|
|