[Elecraft] 5 KHZ dead spots

[wayne burdick]

Merv Schweigert wrote:

> Have noticed this on other bands but paid no particular attention,
> but today was tuning 160 meters and every 5 khz there is a pop
> or change in background receiver noise when tuning.

Hi Merv,

This is normal. The K3 has an incredibly low-noise analog/digital 
synthesizer. To achieve this performance, we have to use a fairly 
low-frequency PLL loop (~22 kHz). This translates into a little 
artifact in the VFO tuning range every 22 kHz, on average, across the 
ham bands. Nearly all other PLL-based rigs have such transitions at 
much wider spacing, because they use up-conversion to a VHF 1st IF and 
thus much higher synthesizer output frequencies. But this results in 
inferior performance, generally, and doesn't support narrow roofing 
filters. (For example, most rigs on the market have a roofing filter in 
the 4 to 15 kHz wide range, compared to as low as 200 Hz in the K3. Our 
down-conversion scheme, to an 8.215 MHz IF, is the reason we can do 
this.)

The tuning artifact is only noticeable if a moderate to strong signal 
is in the passband as you tune across it. When you stop tuning, it's 
gone. And while you're tuning, we do a partial mute of the receiver 
(actually, we engage a low-threshold limiter for about 10 ms). This 
makes it barely noticeable in most cases, which is why, with nearly 
3000 K3s shipped, you've never heard a description of it until now :)

The actual spacing of the tuning artifacts varies on a per-band basis, 
being smaller on the lower bands. The reason for this is a bit 
complicated. We shoe-horn the reference signal for the PLL (i.e. the 
DDS, the source of fine-tuning steps) through a 2.5 kHz crystal filter 
to ensure there are no DDS spurs in the VCO output. 2.5 kHz is a lot 
narrower than 22 kHz. So we have to constantly adjust the PLL divider 
values (N and R dividers, to be precise) so that our DDS signal always 
stays within the crystal filter, yet allows the PLL to generate a wide 
enough frequency range to cover the segment of interest. We used a 
home-brew PC program to pre-calculate the optimal N and R divider 
values for the entire tuning range, and we store them in about 30 kb of 
the front panel flash memory chip. As you tune the VFO, we look up the 
table values.

This works smoothly, but as you noted, you might hear a soft "blip" 
under just the right circumstances as you cross one of the transition 
points. It's a little reminder of what it takes to create a very quiet, 
high-dynamic-range RX.

73,
Wayne
N6KR


---

http://www.elecraft.com