[time-nuts] Re: Dr Kinali - The Thesis

[Attila Kinali]

A wonderful good afternoon everyone!

On Wed, 12 Oct 2022 11:53:17 +0200
Magnus Danielson via time-nuts <time-nuts at lists.febo.com> wrote:

> I just saw the PhD defense of Attila Kinali, which was succsessful and 
> render him being Dr. Kinali.

Thanks everyone for all the wellwishes. 

As promised, I put my thesis online, which you can find at [1].
And of course it is in English... I'm insane but not insane
enough to write in German ;-)

The thesis is split into three parts. 

The first part deals with the Fourier transform of signals that have infinite
energy (i.e. all signals of infinite length). In particular, the Fourier transform
is modified to work with these signals properly. The matching inverse Fourier
transform is introduced. And of course proofs that all this works as one would
expect to. But be warned, the math is dense and the write-up is less than
ideal, to put it mildly.
I then use this modified Fourier transform to define noise and 1/f^a noise
in a way that it works both in time and frequency domain, while having all
the properties we usually attribute with these types of noises.

If you deal with power spectral densities and other types of Fourier transform
type analysis of signals, you should definitely look at this part.
This part was written to correct about a century of EE literature that deals
with infinite energy signals in a mathematically incorrect way. If you have
a textbook that does not explicity handle infinte energy signals or does not
warn that the math does not work for those, chances are that the book has mistakes
in the math, if it isn't outright wrong (guess how I know).  

The second part is basically a summary of the noise propagation in amplifiers
and other non-linear electronics that I have written earlier. If you have not
seen it yet, this is probably the part of my thesis with the most practical
application and impact. In short summary, it explains and extends Enrico
Rubiola's φ-type and x-type noise model and allows accurate noise predictions
in various types of circuits. In particular it explains why all noise models
of amplifiers only ever model additive noise, but noise figures are always
given in dB, i.e. are multiplicative noise.

The third part deals with fault-tolerant clock synchronization and the
metastable effects that occur when you try to do that as quickly as
possible. This part is probably the least interesting to the time-nuts
community. So feel free to skip it.

If you have any comments or question, please don't hesitate to contact me
on or off list.


Greetings and have a nice afternoon/evening

			Attila Kinali
 

[1] "On Time, Time Synchronization and Noise in Time Measurement Systems",
by yours truly, 2022
http://time.kinali.ch/thesis-final-kinali.pdf

-- 
In science if you know what you are doing you should not be doing it.
In engineering if you do not know what you are doing you should not be doing it.
        -- Richard W. Hamming, The Art of Doing Science and Engineering