On May 31, 4:04 am, "François Guillet" wrote in
formatting link
=en :
"GreenXenon" a écrit dans le message de news:
> snipped-for-privacy@b1g2000vbc.googlegroups.com...
> > > "GreenXenon" a écrit dans le message de news:
> > snipped-for-privacy@g19g2000vbi.googlegroups.com...
> > ...
> > | In my hypothetical device the input of a signal that has a frequency
> > | of A Hz and a peak-to-peak amplitude of B volts will result in the
> > | output of a signal that has a frequency of B Hz and a peak-to-peak
> > | amplitude of A x [1.602 × 10^-19 volts].
> > ...
> > Even if the charge is quantified, a potential difference is not.
> | Ok.
> > Why should a pp amplitude be a multiple of 1.602 × 10^-19 volts?
> | Because 1 electron has a charge of 1.602 × 10^-19 coulomb.
> If you are ok with the first statement, you cannot agree the second.
Well, the voltage is not exactly 1.602 × 10^-19 but close. Right?
Let a capacitor C retaining a charge Q. The potential difference is > U=Q/C.
> As C is not quantified, U is not quantified. By changing the plates
> distance, you can adjust C to get any value for U including
> non-multiples of 1.602 × 10^-19 volts. You can have U = 2.1*10-19 V o r
> anything else.
> | There are several applications I can think of for the aforementioned > | device:
> | 1. Transmitting/recording too high a frequency signal on a medium that
> | does not have the bandwidth required to handle the high-frequency
> | 2. Transmitting/recording too large and amplitude signal on a medium
> | that does not have the dynamic range required to handle the large > | amplitude
> | 3. Generating a higher-frequency signal from a bunch of lower- > | frequency signals
> It is theorically feasable. For example, you could do an A/D conversion
> of what you are interested in in your input signal and code the digital
> data with the amplitude of the transmitted signal. Using steps of 10^-18
> volts, a 0->1 V signal could carry in one shot a near 60 bits word!
> Using a 1 Mhz pass-band, you could transmit at least a 60 Mb/s signal.
> Physically it's not possible. The problem is less in the A/D conversion
> than in the noise which avoids to recover the information when a too
> small step is used for the discrimination. Even if no noise was added to
> the signal during its transmission, it is a limit to the initial A/D
> conversion. An "exact" frequency-to-amplitude and visa versa, or
> anything else, cannot have an arbitrarily accuracy due to the noise. The
> noise independantly of its origin (thermal, quantum...) is the only
> limit to signal transmission. See Shannon.
If 1.602 × 10^-19 volt is too small, then what is the smallest physically-possible voltage that can be detected or processed given the state of today's technology?