[HBR] Re: W1KLK receiver

[waltah at earthlink.net]

This is an interesting discussion.

Bill said:
> > Doug's purpose with the elaborate RF stage is to eliminate the image; with 
> > a 455KC IF, you generally need all the front end image rejection you can get.

A 455 kcs IF works fine for an 80-only receiver.   At 3900, the LO 
will be at 4355 and the image at 4810; that's a pretty substantial 
distance away and even one decent-Q RF tuned circuit will take it 
out unless there's a very strong signal there.   However 455 kcs 
was always a problem with multiband receivers on 20M and above, 
frequently also on 40 because of the strong signals from the 
SWBC band in that area.

In the 60's, the region around 5 Mcs was still in use by the military 
for various mostly low-power applications.  It wasn't a bad place to 
have an image response.

On 10M, the less expensive ham receiver designs had two bands 
of equal sensitivity located 910 kcs apart.  
 
Jim said:
> Agreed but I read the article a bit differently.
> 
> I think you could get adequate image rejection without all those
> tuned circuits. What 'KLK wanted was not just image rejection but
> rejection of strong out-of-band signals - meaning anything more
> than a few dozen kHz away from the desired signals. If the signals
> don't reach the mixer, it doesn't have to deal with them. 

Yep.   For 80M, however, he probably could have used the 7360 
mixer 'barefoot' and done even better; there's no serious problem 
coupling two or three tuned circuits without a tube between if you 
are only covering 500 kcs at 80M.  

I don't know what the ultimate sensitivity of the beam tubes is as 
mixers but it's pretty good.   I've got 0.25 uV on 40 meters with no 
RF stage and two tuned circuits preceding the mixer.  For a single 
band receiver you could use a push-pull 6J6 oscillator, with 12AT7 
cascode buffer if needed (probably not) -- the design should be a 
cinch.   With just one oscillator he wouldn't have the problems I 
have with spurious responses.  

The purest would probably use both 'top' coupling for the front end 
coils (with a small cap between the tuned circuits) and 'bottom' 
(with a shared small coil at the bottom so the gain could be leveled 
across the band.   In a bandswitched receiver you run out of space 
pretty fast with so many parts but for a single band it's surely not a 
problem.

An interesting design would be plug in front ends, with different 
VFOs according to the band.  You could even use different 1st IFs 
and convert to a common IF to do most of the amplification in the 
mainframe.   Four tubes, maybe, in the plug-in unit and the rest on 
the mainframe.   

> If I understand the process, we hams measure 3rd order IM by injecting two
> relatively close-spaced signals into the rx and raising the level until an
> inband distortion product shows up. Typically it's two signals 20 kHz on
> 20 meters. 

With practical front end tuned circuits testing at 20 kcs spacing on 
20M tests only the mixer because the practical Q's give you no 
help rejecting the unwanted signals.   On 80M, both the front end 
coils and the mixer are tested.   Both tests are of interest.

That actually works out pretty well because there's usually a much 
bigger problem with IMD on 80M than on 20M. 

On the HBR-4 project, I hope to get 30 and 20M working today.   
The Butler oscillator is actually working reasonably well now, 
though I am going to go crazy with the retuning all the crystals 
every time I move that trimmer board to add another band or two.  
Those 50 mmf trimmers tune so sharply on the high bands that you 
really have to hunt for the crystal frequency.

An additional problem is tuning drift, probably due to heating of the 
toroid.  It usually takes a couple of retunings after each change to 
find the point at which the CO will *keep* working as the set warms 
up.  There are higher stability cores available; I may have to look 
into that.   Or I can just use a slug tuned coil with a VHF core -- 
that's probably simpler.

Walt 
KJ4KV