[Elecraft] Re: Unique synthesizer LO ?

[wayne burdick]

PE1E wrote:
>

> It is said that the synthesized LO of the K2 is of a unique design.
> Though, when I compare the LO phase noise specs of the K2 and the 
> AOR7030,
> the latter seems to show much better specs.

Hi Peter,

Eric (WA6HHQ) and I designed the K2's synthesizer, so I'll try to 
answer your questions.

When we designed the K2 synth we were faced with some difficult 
criteria. Since the rig is a kit, we wanted to keep cost and complexity 
low and avoid using surface-mount parts. And since it's targeted at 
portable operation, we also needed to minimize current drain. But we 
also needed a reasonably high-performance synthesizer to match the 
possibilities of our down-conversion superhet receive architecture.

For all of the above reasons, we decided against using a DDS (direct 
digital synthesizer). "High-performance" DDS chips generally are 
expensive, have high current drain, and require a lot of support 
circuitry. And lately just about all of them are fine-pitch SMDs. In 
1998 when we first started prototyping synthesizers, the choices 
considerably bleaker.

DDS chips have a bigger problem, which for a rig like the K2 was not 
acceptable: close-in spurs that could compromise performance in high 
QRM conditions. These spurs can be managed by adding a lot of 
additional filtering and various techniques that are 
component-intensive. For example, putting a PLL after the DDS will 
help, but without great care, this can actually make things worse (the 
PLL can multiply or fold back in spurs that occur way outside the DDS's 
nominal output frequency). As is clearly shown in QST reviews of radios 
that use DDS without a PLL (e.g., the SDR-1000), a simple low-pass 
filter won't do the trick either. Numerous large spurs can be seen near 
the carrier in phase noise plots of these radios. But a bare DDS is a 
good choice for an "SDR," since it permits the very fast, very accurate 
tuning needed for sophisticated digital modes.

I'm not familiar with the AOR7030's synth design. But if it uses a DDS, 
it must either be managing the spurs as I mentioned above, or it has a 
*lot* of spurs. You won't necessarily see then in the phase-noise plot; 
doing the phase-noise sweeps with a typical bandwidth of 100 Hz won't 
show any but the largest spurs (such as the loop spur shown in the 
AOR7030's plot). But you'll hear them in the form of reciprocal mixing.

Back to the K2. Since we were avoiding DDS, we used a clean VCXO as the 
PLL reference oscillator, tuning over a small range. To tune the 
oscillator, we used a 12-bit DAC driving varactor diodes, and (as you 
may recall) there's a calibration pass where we measure the VCXO and 
store constants to tune it over its full range.

Again, I'm not familiar with the AOR design, but over the range we 
typically measure (carrier + 2 kHz to 20 kHz), their "smoothed" phase 
noise appears to be similar to the K2's. I don't know what bandwidth 
they used for their phase-noise plots, so it's hard to say whether 
they're mitigating DDS spurs or not. As I mentioned, they can be hidden 
by the analyzer's bandwidth. (Note that the upper curve of the two 
shown for the AOR7030 is the one I'm referring to. They provided an 
additional "far out" curve below it, and we don't have a plot of this 
type available for the K2.)

The other piece of the puzzle was the VCO. Somehow, using a minimum of 
parts, we had to make the VCO cover 9 or 10 bands. The resulting 
circuit, with three latching relays to switch fixed capacitors and 
varactors, and only one VCO inductor, did the trick.

73,
Wayne
N6KR

---

http://www.elecraft.com