[hpsdr] Fwd: Horton LO - further thoughts

[Hans Summers]

I think that Jim raises some very interesting points.

I suspect that it will be very hard, if not impossible, to beat the
low phase noise of a decent DDS with a quality crystal reference
oscillator.

The trouble with DDS is as we know, the spurs problem. Even with a
high speed 14-bit DDS such as the AD995x series the spurs can still be
a problem if the rest of the receiver is sensitive enough to notice
(which I think we can assume it is!).

Wideband spurs can be removed by a bandpass filter at the DDS output.
This would require either a tuneable bandpass filter (e.g. by switched
capacitors), or keeping the oscillator range narrow - which could be
accomplished by using a VHF oscillator and a variable division ratio
to get down to the desired HF frequency.

The bandpass filter can't remove close in spurs which also exist and
are just as troublesome. However, the largest source of spurs is the
phase truncation in the DDS and thise location of these spurs is
entirely predictable.

Wouldn't it therefore be possible to say Ok, we want to receive on
frequency X. But if we set the DDS there, we predict we'll get a spur.
So we move the DDS to somewhere within the notional 10kHz of where it
truly should be. And correct for the < 10kHz displacement digitally
within the DSP (as is being proposed anyway).

Wouldn't this be a good way to have the best possible phase noise
using a DDS, and yet to avoid the spurs problem of DDS? I wonder if
this would be a better method than the currently discussed VCO/PLL
type synthesisers?

73 Hans G0UPL
http://www.hanssummers.com


> I'll admit right off that I'm new to ham radio but have done a lot of
> reading in the last year. Although I worked 30 years as an EE before
> retiring I never worked in RF so there's been a lot of fun learning.
>
> The one thing I keep reading is that good phase noise performance in a good
> design is limited by the Q of the resonator. At HF nothing comes close to
> the Q of a crystal. For frequencies up to 500Mhz I'm guessing that it will
> be very hard to beat the phase noise of a frequency multiplied 100Mhz
> oscillator.
>
> Second best is probably a well designed and accoustically isolated resonant
> cavity obviously at a much higher frequency considering practical sizes.
>
> Although the DDS does have SFDR limitations at least the phase noise
> approaches that of a crystal oscillator.
>
> I see mention of fine tuning resolution of the DDS as unnecessary yet it
> comes for free so I can't see it as a limitation. Small frequency changes
> can be useful for minimizing spurs and can be made invisible by software.
> Nothing says you have to use the fine stepping of the DDS for tuning. The
> SFDR is still a limitation.
>
> I guess what I'm saying is that without a good Q at the heart of the
> oscillator I doubt we'll be satisified with phase noise results.
>
> And achieving that maximum base Q may well require different means as the
> frequency goes beyond HF.
>
> jim ab3cv
>
>
>

 1150575032.0