[HBR] Another Receiver Project -- HBR-4, Part 13

[waltah at earthlink.net]

Heading into the home stretch, I think ... Now working on the 
spurious responses.   

There is about one serious spur per band, S9 to S9+20 strength on 
the bands 80 through 20 meters.    Some things that I have 
tinkered with are:

A.   Increasing the oscillator cathode resistor.   This seems to 
help, presumably by reducing the osc. grid current and hence 
distortion.  

B.  Reducing the VFO output.   This helps but there's a trade-off 
with sensitivity.  

C.  A 9 Mcs trap in the premixer plate circuit.   The strong spurs all 
seem to leave the premixer at 9 Mcs so this helps considerably.   
It is complicated by the push-pull circuitry and there's interaction 
between the trap and the tuning of the premixer plate.

A possibility that I haven't tried is a high pass filter between the 
premixer and the mixer, cutoff above 9 Mcs but below 16 Mcs.   
This wouldn't be as effective as a 9 Mcs trap because the strong 
spurs all involve 9 Mcs.  

A low pass filter at the output of the VFO to the premixer should be 
helpful since all significant spurs involve a 3rd harmonic or higher 
and much of the harmonic generation is probably in the oscillator.  

Further reducing the oscillator grid current should help the spurs 
involving an even harmonic of the VFO.   The odd VFO harmonics 
are more likely the result of imbalance in the premixer with VFO 
imbalance a secondary issue.

What about jiggering the VFO and crystal frequencies slightly to 
(possibly) move the spurs out of band?  The worse ones on the first 
four bands are:

(dial freq -- strength -- multiple x osc. freq = freq - xtal = 9 Mcs)

3600 -- S9+20 -- 5 x 5.4 -> 27 - 18 = 9 Mcs.

7417 -- S9 -- 6 x 5.083 -> 30.5 - 21.5 = 9 Mcs

10,334 -- S9+20 -- 3 x 5.166 -> 15.5 - 24.5 = 9 Mcs

14,000 -- S5 -- 12 x 5.5 -> 66 - 2 x 28.5 = 9 Mcs

14,143 -- S9 -- 7 x 5.357 -> 37.5 - 28.5 = 9 Mcs

14,167 -- S9 --  9 x 5.333 -> 48 - 2 x 28.5 = 9 Mcs

The rest are all at the S0 level -- you can hear them with no 
antenna connected but any significant noise or signal would cover 
them up.

Note that those that are in band are near the low end.   This 
suggests tweeking the VFO and crystal frequencies slightly 
downward.   

For example moving the VFO range down 125 kcs (4.875 - 5.375) 
and the crystals down by the same amount gives us:

(dial freq -- multiple x osc. freq -> freq - xtal = 9 Mcs

3500 -- 5 x 5.375 -> 26.875 - 17.875 = 9 Mcs

And thencely:

7312 -- 6 x 5.063 -> 30.375 - 21.375 = 9 Mcs

10,250 -- 3 x 5.125 -> 15.375 - 24.375 = 9 Mcs

13,896 -- 12 x 5.5 -> 65.75 - 2 x 28.375 = 9 Mcs

14,036 -- 7 x 5.339 -> 37.375 - 28.375 = 9 Mcs

14,069 -- 9 x 5.306 -> 47.75 - 2 x 28.375 = 9 Mcs

That's better but not what you'd call a cure, especially since there 
are also spurs on the higher bands.

It seems that the first thing is to optimize the VFO operating 
conditions.  It should be as close to class 'A' operation as possible 
because grid current drawn on positive peaks causes even-
harmonic distortion of the waveform.  I'll look at the oscillator coil 
and see if the Q can be increased; unfortunately that would require 
a heavier gauge wire and there probably isn't space on the form. 

With an oscillator signal in the 10V p-p range and spurs of perhaps 
100 uV, it doesn't take a lot of distortion!  

Second to that, the premixer needs to be accurately balanced, with 
the adjustment made while watching an odd-harmonic spur.   
Some further improvement may be gotten by selecting the 
oscillator tube for minimum odd harmonic output -- the halves are 
never perfectly balanced.

Then start with the traps and filters.  More to come, obviously.

Walt
KJ4KV