Hi Guido and all,

> Wonder if thresholding a sine wave is increasing the phase noise. When
> thresholded the period jitter is soley based on a single instance of the
> amplitude transition, while phase errors during for the full sine period
> could average out period jitter. Or is this not the case?
Phase noise is typically a problem close to the carrier but of no
importance at large frequency separations.

Whenever you transform a sinewave into a squarewave you will add
wideband phase noise, but under normal circumstances that is of
no importance because narrowband phase noise close to the carrier
is so much stronger. The narrowband phase noise corresponds to a
systematic time jitter. The sine wave is phase shifted for quite 
some time.

The original thread, why not use very low frequencies, has the 
conventional answer that we have to avoid the 1/f noise and the 
interference from 50/60 Hz hum. A good design can make these problems
insignificant, but another problem is that soundcards are not DC
coupled. A card like Delta 44 (which still is my favourite) does not
only use a capacitor to block DC at the input. I could easily 
eliminate that, but the chip used in the card also has a "DC 
block in software," a digital high pass filter that ensures 
that the average of a very large number of samples always is zero. 
That is a high pass filter that prevents a DC offset on the 
input to reach the SDR software and therefore causes an infinitely
deep notch at the center.

With a good design the noise near the passband center can be made 
negligible. See for example
http://www.sm5bsz.com/digdynam/practical.htm
Figures 5 to 10 show the center 1/f noise of the Delta 44
becoming less significant as the input level is increased
and the noise from the antenna gradually hides the "center
spur" Figures 11 to 13 show that the 1/f noise is totally
negligible when the system gain is adequate.

There is a lot to say about the proper design of local
oscillators (and soundcard modifications) to avoid the
low frequency noise in direct conversion receivers. Avoiding
AM sideband noise is particularly important when schottky
diode mixers are used to produce I and Q.

73

Leif / SM5BSZ


Reply via email to