[Collins] Any Frequency

[Dr. Gerald N. Johnson]

On Tue, 2009-10-06 at 23:46 +0000, Kees & Sandy wrote:
> I ask you to please check the results of testing done by Jack Smith of
> Clifton Labs and check the specs SiLabs has on the part. The lack of
> spurs on the Si570 output relative to very frequency agile DDS chips
> is very impressive. It's as good/better than the HP 8640 signal
> generators.
> 
> 73 Kees K5BCQ 
> 
As I said, Smith's testing ignored the basic problem of a DDS. Because
the output sinewave is generated by discrete steps at the sampling rate
(internal or external clock) there is noise from those steps. Its
effectively a beat note between the harmonics the output signal and the
sampling clock. When that beat note is a exactly zero, the output is
very clean. Otherwise there is a comb of spurs at the beat frequency
intervals. This is documented well in the ARRL book, Experimental RF
Circuit Design, the last page of section 4.6.

When the output is a square wave the crud spectrum is unbelievable. I
found that at least 30 years ago when I thought I didn't need to bother
with the sine wave making, that I could take off the most significant
bit of the accumulator of a DDS.

You don't need a fancy spectrum analyzer to find those spurs. Your
receiver will do. Set the DDS to a convenient frequency, and hook it to
your receiver in place of the antenna. Attenuate it so it doesn't quite
overload the receiver. Tune away from the signal frequency in CW or SSB
mode. You will see the phase noise close in, probably out beyond 100 KHz
with the conventional Leeson slope (10 dB per decade of offset) and once
you get out beyond most of the phase noise you WILL find a forest of
spurs. Change the DDS frequency and you fill find the forest has moved,
maybe increased in amplitude, maybe decreased in amplitude.

More bits in the accumulator reduces the amplitude of the spurs but
can't fundamentally prevent them for all possible output frequencies.

So I'm very skeptical.
-- 
73, Jerry, K0CQ, Technical Advisor to the CRA
All content copyright Dr. Gerald N. Johnson, electrical engineer