Reply-To: "Kevin Aylward"
From: "Kevin Aylward"
Newsgroups: sci.electronics.design
References: <4c3c095a.0212231022.fdd6810@posting.google.com> <85fg0vc4obu9t3plhdilkbk0i564mafv4e@4ax.com>
Subject: Re: what is monte caro analysis ?
Organization: AnaSoft
X-Newsreader: Microsoft Outlook Express 6.00.2800.1106
X-Inktomi-Trace: public1-pete2-5-cust66.pete.broadband.ntl.com 1040769889 9623 80.0.216.66 (24 Dec 2002 22:44:49 GMT)
Message-ID:
Date: Tue, 24 Dec 2002 22:44:49 -0000
NNTP-Posting-Host: 80.1.224.4
NNTP-Posting-Date: Tue, 24 Dec 2002 22:44:49 GMT
R. Lewis wrote:
> "Paul Burridge" wrote in message
> news:85fg0vc4obu9t3plhdilkbk0i564mafv4e@4ax.com...
>> On Mon, 23 Dec 2002 22:41:45 -0000, "Kevin Aylward"
>> opined thusly:
>>
>>> boki wrote:
>>>> Hello All:
>>>>
>>>> I have to test temperature and power supply by monte caro analysis,
>>>> but I don't know what is monte caro analysis ??
>>>>
>>>
>>> This is where all component values are randomly varied.
>>
>> Wouldn't it be more useful if the values were instead changed to
>> their maximum/minimum tolerances?
>> --
>
> The reason for a Monte Carlo is that the max/min tolerances are seldom
> responsible for the worst case scenario where any reactive
> components/traces are around.
Its certainly true that WC does not *guarantee* a wc output condition.
However, I think the use of the word "seldom" is really stretching it a
bit for *general* circuits. I would estimate that in probably 80%+,
maybe even 95% of cases WC will give an estimate that is actually worse
than the wc would be in practise. This is because, in part, some
parameters have inverse correlations.
In practice, it is not usually possible to do a decent MC at all for
transient runs, because it simple takes way to long.
Kevin Aylward
sales@anasoft.co.uk
http://www.anasoft.co.uk
SuperSpice, a very affordable Mixed-Mode
Windows Simulator with Schematic Capture,
Waveform Display, FFT's and Filter Design.