home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!tymix!tardis!olivea!decwrl!ames!data.nas.nasa.gov!taligent!apple!netcomsv!attain!icd.teradyne.com!news
- From: eisenhof@zax.eda.teradyne.com (Karl Eisenhofer)
- Newsgroups: rec.audio
- Subject: Re: Where can I find the Nyquest theorem
- Message-ID: <1992Sep1.200744.19701@icd.teradyne.com>
- Date: 1 Sep 92 20:07:44 GMT
- References: <1992Aug25.234452.4677@ptcburp.ptcbu.oz.au> <khaw.714892058@parcplace.com> <1992Aug31.114555.17504@discus.technion.ac.il>
- Sender: news@icd.teradyne.com (News admin)
- Reply-To: eisenhofer@maven.dnet.teradyne.com
- Organization: Teradyne, Inc. Boston MA
- Lines: 22
-
- In article <1992Aug31.114555.17504@discus.technion.ac.il> sorin@techunix.technion.ac.il (Goldenberg Sorin) writes:
- >Question
- >A signal say x(t) is bandlimited to W hz.
- >I produce a signal x'(t) by multiplying the original signal x(t) by 0 at
- >times i/2/W. x'(t)'s spectrum is the same as x(t) s. Then I sample x'(t)
- >at time i/W/2, so I get a sampling frequency of 2W. All the samples will
- >be 0, so I can't construct x'(t) from it's samples. What went wrong ?
-
- What went wrong is that x'(t)'s spectrum is NOT the same as x(t)'s
- spectrum. In fact, the multiplication step introduced a whole slew of
- high frequency components (all the way up to infinite frequencies)
- into the signal. Now this signal is no longer bandlimited and the
- Nyquist theorem does not apply. The samples that you get will construct
- and alias of the signal (in this case, an alias with frequency 0).
-
- I am doing this from memory, no textbooks or anything, so please
- flame gently (also, this is not my area of expertise).
-
- Karl
- --
- Karl Eisenhofer SPIKE eisenhofer@maven.dnet.teradyne.com
- "Searchlight casting for faults in the clouds of delusion"
-