# staticstical fluctuations

Discussion in 'Electronic Basics' started by PZ, Jan 12, 2006.

1. ### PZGuest

Quote:

" As the number of electrons N decreases, the statistical fluctualtions
in the number becomes an increasing fraction of the total, limiting
circuit performance and making circuit design more difficult.

".

this doesn't prevent me from further reading, just wonder what
statistical fluctualtions actually mean, in simple language.

thanks

2. ### Charles SchulerGuest

A basic example would be to measure the height of 100 people. You will get
100 different values (assuming your method of measurement has enough
resolution).

A second example would be to measure the output voltage of 100 photo sensors
that were illuminated with a constant source. Again, 100 different values
are expected.

A third example would be to collect data on the time before failure of 100
light bulbs.

Statistics are used to deal with data that are known to be subject to random
fluctuations. Almost all data are, by the way.

3. ### redbellyGuest

Think of the current as a measure of how many electrons pass through a
wire every second.

But, this number is just an average. If you were to count the actual
number of electrons every second, it would usually be different than
the average number. These fluctuations from the average are what the
person you are quoting is talking about.

Mark

4. ### Rich WebbGuest

The bigger your sample is, the closer your estimate of [something] is to
the estimate that you would make with subsequent large samples.

You're probably most accustomed to [something] being the average of some
measurement of the sample, so consider an average weight. If you take
the average weight of samples of, say, five people at a time then the
results depend a lot on which five people you choose this time as
opposed to next time, and your sample to sample variation will be large.

If, on the other hand, you weight people in groups of 500 then you would
intuitively expect that the average weights of successive groups would
be closer to each other than the average weights of the groups with just
five people each. (For this statistic, it turns out that the expected
value of the spread of the average will scale by the square root of the
number of samples but that's not really germane here.)

So, turning it around, starting with mean weights determined by samples
from large groups, the fluctuation in the results will be greater when
the number of samples include in each group is reduced.  