[Elecraft] IF Shift, PB Tuning, and why neither completely eliminates QRM

[wayne burdick]

These terms are often used interchangeably. It's like the terms 
"biannual" and "semi-annual." One is supposed to mean "every six 
months" and the other "every other year," but some dictionaries list 
them as synonyms, because you can make a case for either word having 
either definition.

Similarly, there's nothing about either the term "I.F. Shift" or 
"Passband Tuning" that helps the average operator tell which one you're 
talking about. They're too vague. But there really are two different 
concepts here. I'll explain what I think they are, and then explain why 
most implementations don't really protect you against out-of-band 
signals.

There are at least two kinds of "shifting" covered by these terms:

- In one case you're just shifting the pitch of the I.F. without 
changing the bandwidth. A more accurate term would be "Pitch Shifting."

- In the other case you're shifting *two* I.F.s in opposite directions, 
reducing the degree of I.F. filter overlap from 100% to something 
lower, so that the ultimate bandwidth (at audio) is reduced. The center 
pitch of the passband may or may not change at the same time. Let's 
call this "Filter Shifting."

Either approach may appear to reduce QRM. But the QRM may in fact still 
be there. How can this be? It's because the bandwidth first filter has 
not actually changed; you're just listening to the portion of its 
passband that's making it through to the second I.F. The rest of the 
first I.F. filter's passband is still there, exposing subsequent stages 
to overload, AGC pumping, etc. That's why some seemingly modern radios 
break down in contest situations: they use a fairly wide first filter 
(often without telling you), then tighten or shift the *second* I.F. 
filter, which you perceive as a reduced bandwidth. But if a signal 
within the first passband exceeds the signal-handling capability of a 
subsequent stage, it sounds like, well--crap. (That's another name for 
serious in-band IMD.)

A much better approach to the "Filter Shifing" problem would be to 
actually *narrow* the first crystal filter, protecting all subsequent 
stages. This is why the K2's variable-passband filter is such a great 
tool. You can narrow it down at the same time you narrow the DSP or 
audio filter. Of course the variable-passband filter is most useful in 
CW mode because it's optimized for narrow bandwidths, but it can 
certainly be used in SSB modes as well.

In the case of the K2, ignoring audio filtering, the "quantized" 
version of I.F. shift that I mentioned earlier (using different BFO 
settings with the same crystal filter) is like the "Pitch Shifting" 
case. However, as soon as the pitch-shifted passband starts to hit a 
wall (namely the basic audio response of the radio and/or the DSP 
filter passband), the bandwidth starts getting narrower, too. This has 
some of the effects of "Filter Shifting," whether intended or not.

Ideally you'd have variable-passband filters usable in all modes, along 
with several controls, e.g. SHIFT, WIDTH, HI CUT, and LO CUT. It would 
then be possible to optimize the passband intuitively, as required to 
deal with QRM (that's the point). And like I mentioned, you'd like to 
have the benefit of not just shifted filters, but filters that get 
narrower as required.

How does all this apply to the K2?

Obviously there isn't room for a suite of passband controls. But you 
might be able to emulate this functionality using a computer, given the 
proper remote control commands. (Sounds like something else for the 
wish-list.)

You could also modify the K2's variable-passband CW filter to optimize 
for wide bandwidths (SSB/DATA) rather than narrow. The same crystals 
are used both in the main (stock) filter and the KSB2. All you'd need 
to do is use smaller varactor diodes, preferably a matched set. You 
might be able to get a 1200-2400 Hz passband range with acceptably low 
ripple.

Food for thought.

73,
Wayne
N6KR


---

http://www.elecraft.com