[Elecraft] Elecraft technology

[Bill Coleman]

On 9/4/03 12:40 PM, Sverre Holm at la3za@qsl.net wrote:

>I received an interesting comment that it is basically the architecture
>of the K2 which is unique for a ham receiver, i.e. the single-conversion
>design and the narrow ham-band filters at the RF input -
>http://www.elecraft.com/Apps/why_is_the_k2_receiver_single.htm, and I
>agree this is key to the K2's performance. I could rephrase my question
>differently, why did it take until 1998 for this architecture to make it
>into ham receivers, why didn't it happen say ten years earlier?

Sorry to answer such an ancient e-mail, but I think there's something to 
look at here.

Back in the 70s or so, there were many tranceivers (or receivers) that 
used a dual-conversion architecture. Typically this would involve 
conversion to a broad-band IF from a fixed heterodyne oscillator, then 
conversion through a 5-5.5 MHz LMO to a crystal filter for the IF. 
Sometimes there was a third IF frequency, but this dual conversion seemed 
typical.

The dynamic range on this architecture wasn't fabulous, but it was stable 
and reliable. Since the LMO only covered 0.5 MHz, a fixed heterodyne 
crystal was needed for each band (or 4 for the full 10m band). On some 
rigs, the LMO was even more narrow, so more expensive crystals were 
required.

Then in the 80s, a rash of solid-state rigs hit the market with 
general-coverage receivers. This required a different approach. Instead 
of an LMO, PLLs were used to generate the conversion frequencies. This 
got rid of a lot of expensive heterodyne crystals. To make the rig 
general coverage, the first broad-band IF was moved up to the 45-70 MHz 
range, and was broader than the 0.5 MHz of before. This worsened the 
dynamic range situation a bit, but the worst part of this change had to 
do with the PLLs. 

Those early PLLs generated a lot of low-level broadband noise, called 
Phase Noise. They didn't generate pure frequencies like those old crystal 
heterodyne oscillators. This meant these rigs received (and transmitted!) 
a low-level broadband response. For weak signals, this didn't matter 
much, but it was terrible for adjacent, strong signals to leach through. 

Many of these rigs also supported FM as a mode, so the first roofing 
filter was typically 15 kHz or more. 

Designers have since cleaned up their PLLs quite a bit. But the middle 
90s saw the introduction of the Ten-Tec Omni V. It solved the phase noise 
problem by going back to crystal-controlled heterodyne oscillators. Like 
its predecessors, it isn't general coverage. But the Omni V showed it was 
possible to get better performance by going back to earlier designs.

In one way, the K2 goes back to even earlier, single-band superhet 
design. That's why its dynamic range figures are so good. Most rig 
manufacturers aren't likely to move back to this simpler design, largely 
because they don't want to give up general coverage receive.

Ten-Tec has come up with a new rig, the Orion, which combines techniques 
of single conversion to a narrow IF and multiple conversions. They use a 
combination of roofing filters which are significantly narrower than 15 
kHz in many modes.

The Orion design works, but it is expensive.

The K2 receiver performance is right up there, still. 

My friend Gary, K9AY told me a couple of years ago, "The K2 is a simple 
radio superbly executed." There's something to be said for simplicity.


Bill Coleman, AA4LR, PP-ASEL        Mail: aa4lr@arrl.net
Quote: "Not within a thousand years will man ever fly!"
            -- Wilbur Wright, 1901