Re: Noise Source with Limiter ?
From: Nicholas O. Lindan (see_at_sig.com)
Date: 10/16/04
- Next message: Fred Bloggs: "Re: OT: Failure to communicate? Genesis"
- Previous message: Fred Bloggs: "Re: APD bias"
- In reply to: Phil Hobbs: "Re: Noise Source with Limiter ?"
- Next in thread: Phil Hobbs: "Re: Noise Source with Limiter ?"
- Reply: Phil Hobbs: "Re: Noise Source with Limiter ?"
- Messages sorted by: [ date ] [ thread ]
Date: Sat, 16 Oct 2004 18:32:42 GMT
"Phil Hobbs" <pcdhSpamMeSenseless@us.ibm.com> wrote
> "Gaussian" means that samples of the instantaneous noise signal form a
> random process whose probability density is Gaussian--i.e. that a
> histogram of the amplitude has a round top and wings that die off
> rapidly but that have no sharp cutoff.
>
> A good limiter produces flat-topped pulses of uniform height, with
> slightly sloping sides. This produces a histogram that has two huge
> peaks at the positive and negative clip levels, plus a low, flattish
> region in between corresponding to the nearly-straight sloping sides.
>
> - ^ ^
> / \ | | | |
> _______/ \_____ vs ______| +--------+ |_______
>
> Not very close at all.
But the destination of most noise is a series of 1's and 0's,
so it has to go through a limiter/quantizer somewhere.
Logically there is no requirement that the prime source of
the noise be gausian.
It is a very tough job to create very good physical 1-0 noise,
at the limit of perfection it is impossible.
It is easy to check for bias in 1-0 noise: keep track of the
ones Vs zeros and the incidence of strings of 1, 2, 3,...
successive zeros or ones. Within a minute, hour, day, week -
dependant on how much spare time you have on your hands to
fiddle with circuit design - a bias shows up.
Von Neuman postulated one can make a perfectly fair coin by
counting a head every time a tail was followed by a head, and
a tail on every time a head was followed by a tail; all strings
of heads or tails are discarded.
If you try Von Neuman's method on physical noise it will fail.
And if you follow one VN algorithm by another it will still fail ...
The coin (zener/astable ...) appears to have an infinite memory.
The basic premise of a random source: that the next value can not
be predicted by past values can not be realized.
Then there is a conundrum asking what is the probability that VN's
algorithm only _appears_ to be false...
Somewhere in an infinite random string there are the complete works
of Shakespeare (and the complete works with one spelling error, and
with two spelling errors...). If there aren't then the source is
not random. It took all the random events of the universe 14.5
billion years to create same, even with one spelling error, and with
two ....
* * *
There are all sorts of things to wonder about at 2am:
An irrational number is by definition both random - there is no
repetition in the digits - and predetermined: Viz. the algorithm to
determine pi is smaller than pi, or pi is compressible.
Now, do the digits of pi contain the works of Shakespeare, how
about Keats ...?
-- Nicholas O. Lindan, Cleveland, Ohio Consulting Engineer: Electronics; Informatics; Photonics. Remove spaces etc. to reply: n o lindan at net com dot com psst.. want to buy an f-stop timer? nolindan.com/da/fstop/
- Next message: Fred Bloggs: "Re: OT: Failure to communicate? Genesis"
- Previous message: Fred Bloggs: "Re: APD bias"
- In reply to: Phil Hobbs: "Re: Noise Source with Limiter ?"
- Next in thread: Phil Hobbs: "Re: Noise Source with Limiter ?"
- Reply: Phil Hobbs: "Re: Noise Source with Limiter ?"
- Messages sorted by: [ date ] [ thread ]
Relevant Pages
|
