[HBR] Yet Another HBR Concept

[hbr@mailman.qth.net]

Making some progress.   Here's another conversion scheme:

VFO tunes 8.7-9.2 Mcs.   11th harmonic of 1.75 Mcs is 19.25 Mcs.
19.25 - 8.7-9.2 Mcs = 10.55 - 10.05 Mcs.   With an IF of 6.55 Mcs 
that tunes 3.5 - 4.0 Mcs.   6.5536 is available as a clock crystal and 
the frequency seems to be clear of high power SWBC stations.   
VFO is above IF so no problems there, all the harmonics of the 
oscillators are well above the injection frequency ... does anyone see 
any problems?

The virtue of 8.7-9.2 Mcs as a VFO range is that's the FT-101 VFO 
and I have another one of those units.   In the original both gangs are 
used with a small coupling cap to tune a single ended oscillator but 
for the HBR2K I used just one gang and had no serious trouble 
getting it to track the dial; I could use both gangs for a push-pull 
oscillator and still get tracking.   

This VFO and dial mechanism is ideal for a really high-quality 
receiver because it's a unit oscillator and the dial is calibrated 
directly in one kcs intervals; marked 0-500 and 500-1000 so all the 
meaningful bands are covered.   The rest of the FT-101 doesn't work 
too well for this new project because the push-pull front end may not 
fit -- but I guess I'll take a look.  I do have another junker chassis and 
a blank top plate.  Secondly, the mounting of the oscillator assembly 
doesn't allow proper cooling for a vacuum tube design -- one 
disadvantage I faced in the HBR2K.   Again, it's worth a look.

Those oscillators turn up on eBay every month or so; they go for $10-
25.   You have a good tuning cap (aluminum plates -- it could be 
better) and a far better gear drive and dial mechanism than any of the 
usual HB projects of the 60's.  Tuning rate about 15 kcs/knob 
revolution with a single stage JB reduction drive between the knob 
and the dial.  

> One thing to consider with the Hahnel oscillator: Locked oscillators 
> of any kind can have significant phase noise. 

Well, yes ... sort of.

A phase locked oscillator stays on a certain frequency because it's 
phase is regularly yanked over to be where it would be (in the cycle) 
if it had been on the right frequency.   But those adjustments cause 
distortion of the waveform.   'Distortion of the waveform' means 'other 
frequencies' -- the yanking process generates noise.   That noise can 
be serious -- and many of the early phase-locked-loop transceivers 
not only had noisy receivers because of phase noise, but they 
transmitted the $#!^ too.   

You can't get rid of PLL noise by filtering because it's worst right next 
to the desired output frequency.   Any filter both narrow enough and 
stable enough to help, would be ideal for use in an oscillator that 
would make the PLL unnecessary!   

Various techniques can be used to make things better but most of 
them apply only to PLLs -- not to a simple locked oscillator like the 
Hahnel idea.   The one thing that does apply is to build a good stable 
oscillator to start with, because the noise is in proportion to how 
much correction is required.   The problem is considerably simplified 
by the fact that we're not talking about a tunable oscillator -- just 
about one that works on a small number of fixed frequencies 17.5 
Mcs (160M), 19.25 Mcs (80), 22.75 Mcs (40) ... that we want to have 
precisely right, rather than just within a few kcs.

Moreover, the noise of a locked oscillator is a better kind.   In a PLL, 
the oscillator 'coasts,' gradually getting further out of phase until it is 
pulled back; how often it is pulled is a design property of the PLL and 
is related to questions of stability and response time of the locking 
loop and there's noise in the 'how often,' as well.  In a locked 
oscillator, OTOH, correction is applied every nth cycle.   For example 
in the proposed design, 80 meters uses the 11th multiple of 1.75 
Mcs so every 11th cycle, the phase of the 19.25 Mcs oscillator is 
corrected by the signal from the 1.75 Mcs oscillator.   The 'noise' is 
actually sidebands that appear at multiples of 1.75 Mcs on both 
sides of 19.25 Mcs.   *These can be filtered*, although at the highest 
multiples -- somewhere around 50 Mcs for 10 meters -- you need a 
pretty narrow filter to get much help.

I need to look at the spurs for the other bands.   Other than that, it 
looks okay.

> ZL2AMJ  solved this problem by using an inexpensive surplus xtal at
> about 3500 kc. on its third and fifth overtones to get output on
> about 10500 and 17500 without even-order harmonics or the
> fundamental making spurs in the front end ...

Slight disadvantages of the method are that it further constrains the 
problem -- wouldn't work with the conversion scheme I'm looking at, 
for example, because I use an odd multiple of 1.75 Mcs.   Also, 
overtones are not precise multiples of the fundamental -- they are 
*mechanical* harmonics and if you look at what the crystal is doing, 
the acoustic standing waves inside the thing have different end 
conditions than those that terminate at the surfaces.   The 
frequencies are typically within a few kcs of the multiple, I think -- 
you can cover by recalibrating the VFO on each band, as would be 
required with separate crystals.   It is, however, a system of 
noteworthy simplicity -- one triode with various non-critical tuned 
circuits to select the overtone is all it takes.

One small virtue of the Hahnel scheme is that once the control 
crystal is precisely on 1.75 Mcs, the calibration is identical on all 
bands  -- just calibrate the VFO and you're done.   A disadvantage is 
that the 1.75 Mcs crystal will put a spur (a marker!) at the lower edge 
of each band.   It's weak on the 1MHBR, but in that one the crystal 
oscillator is injected into the 2nd mixer -- in premixed form, it'll be 
right up there with the RF stage.

Regarding the W5OMX receiver:

> Note that the premixer is used on all bands even though 80 and 20
> can be tuned directly by the 5-5.5 VFO output. Why this was done is a
> mystery, because it has to be a source of avoidable birdies and
> spurs. 

Doing otherwise gives you the problem of matching gains and 
assorted possiblities coming from having a different signal path on 
those bands.   

Thanks for the info on the W5OMX receiver ... now *there's* a classic 
set you'd like to see turn up under a table at a fleamarket.   Maybe 
even on eBay -- how many hams would recognize it?   The call 
appears to be unassigned at present.   

Walt 
KJ4KV