[time-nuts] Building a DMTD/phase noise set in the 21st century

[Attila Kinali]

On Sun, 25 Aug 2019 13:59:15 +0200
Gerhard Hoffmann <dk4xp at arcor.de> wrote:

> Am 25.08.19 um 10:52 schrieb Hal Murray:
> > attila at kinali.ch said:
> >> As you have always two channels, I would recommend to use two mixers fed with
> >> an LO that is 90°C out of phase to get I and Q components.
> > How does that compare with running a single channel twice as fast?

> At low enough frequencies, the two ways are identical in performance
> and linked by the Hilbert transform.

Yes, exactly. This is the ideal case where it doesn't matter whether
one takes n-times more samples per second by sampling faster or by
using n-times as many ADCs. The I/Q sampling is only meant as a way
to possibly decorelate noise that is not inherent in the signal.
 
> At high enough frequencies, ADC performance starts to suffer. One gets
> a smaller "effective number of bits"  ENOB  as sample rate and/or input
> frequency rise.

This starts quite low already, unfortunately. There are several sweet spots
where SNR vs sampling speed is optimal. These happen usually when a certain
ADC technology comes to its limits. Fortunately, these limits, and the
associated sweet spots are moving upwards as time and technology progresses.

A few years back, when we designed the sine-exitacion based TIC[1],
the sweet spot for pipeline ADCs was somewhere around 100-120MHz.

				Attila Kinali

[1] https://ohwr.org/projects/r19-tdc-del-a/wiki
-- 
It is upon moral qualities that a society is ultimately founded. All 
the prosperity and technological sophistication in the world is of no 
use without that foundation.
                 -- Miss Matheson, The Diamond Age, Neal Stephenson