[HBR] HB-67

[hbr@mailman.qth.net]

Jim made many excellent comments, beginning with:

> It would be interesting to know how good or bad the HB-67 really is.

Yep.  I said what I think, but measurements are a lot better than 
theory.   

> There does need to be enough gain to make up for the losses in the
> input tuned circuit and mech filter, though. Also, having adequate
> gain means no RF stage is needed. Finally, the front end has to be
> gainy enough that the first rx stage and not the first if stage noise
> is the dominant factor. 

All of this reduces to 'enough signal/input noise exiting the filter to 
cover up the 1st IF stage noise.'   But on 75 meters this is not much 
of a challenge because the atmospheric noise level on the band is so 
high.   No reasonable mixer on this band requires any preceding gain 
at all -- any triode mixer (for example) will work fine with no RF stage, 
indeed, probably any decent pentode would deliver an adequate 
noise performance.  On higher bands the HB-67 adds a converter and 
the properties of the converter then dominate noise considerations -- 
but with a high gain second mixer (the 7360) you're looking at a great 
deal of gain prior to the filter.  

The highest performance homebrew receiver design I know of is the  
G2DAF Mk II receiver in the RSGB *Handbook*, 4th edition.   The RF 
stage is a push-pull 7ES8, AGC controlled and always operated at 
less than unity gain.  The 1st mixer is a ECC88 (6DJ8), again, push-
pull, driven by a push-pull crystal oscillator to convert to a tunable IF 
5-5.5 Mcs.  The 2nd mixer (still push-pull, an ECC85, 6AQ8) gets 
single end input from a VFO, converting to a fixed IF of 455 kcs; the 
selectivity is provided by a mechanical filter.   Claimed dynamic 
range 100 db; claimed sensitivity <0.5 uV for 10 db s/n on all bands.  
I know of no reason the claims should be doubted.   

On 10 meters and above one can probably improve performance with 
a low noise RF stage of modestly over unity gain.  Below there an RF 
stage is at best an AGC-controlled attenuator and at worst a serious 
contributor to mixer/filter overload.   Of course this assumes 1960s or 
later tubes.   In the 1940s, the picture was different (6K8s for mixers 
...) and an RF stage would have been most helpful.   

The 1MHBR got an RF stage because I wanted to keep things 
simple by distributing the gain over various frequencies and in a 
design of this type I was willing to accept some degradation of the 
dynamic range.  The HBR2K got an RF stage because I wanted 
someplace to apply AGC ahead of the 1st mixer and because it 
simplified using the existing front-end coils; the jury is still out on 
whether >80 db dynamic range can be obtained.  

In a design 'from a blank sheet of paper' and aiming at top 
performance, I'd follow G2DAF on the RF stage and 1st mixer.

> I always wondered why they did that, instead of another 7360. The 
> 6BA6/6D10 converter did not save money, space or milliamps.

Considering the casualness of the original design, I think it was just 
a mistake.  

>> I believe that the later W6TC HBR designs will give better 
>> performance.    

> All depends what you mean by "better". The HBR designs can be drifty
> because of the high frequency LO. The selectivity in a W6TC rx is so
> far from the antenna (3 stages until the first 100 kc. IF can, and
> it's not the entire knothole) that dynamic range has to be
> comrpomised compared to a single-conversion-one stage-before-
> the-filter design.

These are valid criticisms.  But the HBR-series are not 'spare no 
effort' designs -- they were intended to give *good* performance (i.e. 
better than the average ham could have afforded in a commercial 
receiver at the time) in a home brew set that was fairly simple to 
duplicate and could be completed in steps, one band at a time.   The 
HB-67 is if anything a more casual design -- it is possessed of less 
design integrity.  While it ought to have less warm up drift on the 
higher bands (for equal care in construction) than the HBR-series, I'd 
expect it to be a weaker performer in almost every other department.  
Note that there's no buffer between the triode oscillator and the 
mixer; a 7360 probably has less coupling between the beam and one 
deflection plate than most mixers do between two grids but I wouldn't 
assume that pulling will not be a problem.   And when you build the 
necessary converter to get multi-band operation, it's not exactly 
simple, either.   

Regarding balanced drive to the beam tube mixer:

> Or better yet, use a phase splitter/isolator tube to get balanced
> drive without all the complexity ... 

That's not exactly a way to simplify things.   A push-pull oscillator 
inherently gives you two signals 180 degrees out of phase -- phase 
balance is guaranteed by the physics of oscillation.  Thus you only 
need to obtain amplitude balance -- a fairly simple operation.   If you 
go to a phase splitter the cathode and plate signals won't be in 
phase because they'll have different capacitive loads.  Okay, so you 
fix that ... but the gain is different to the plate compared to the 
cathode so the phase is *still* not quite right ... then you've got the 
amplitude to deal with.   For narrow bands, you can probably work 
out a two-adjustment scheme (per band) that will be satisfactory but 
that's a bunch more work than just using a push-pull oscillator.

As to the rest of your proposed design, I generally agree.  Certainly 
premixing a tunable oscillator with crystals to get the desired band 
and thus eliminating the second mixer on the signal path should 
simplify the overload situation -- an amplifier is inherently more linear 
than most practical vacuum tube mixers.   Just a few comments:

1.  Premixing trades gives you more spurious response issues in 
exchange for the reduced overload problems.   That's probably a 
good trade-off for a ham design.  It also increases complexity 
significantly because the premixer output must be tuned across each 
band -- or broadbanded.   The added complexity is an issue in a 
homebrew design.   

The trade-off really should be explored.   I don't know of a well-worked-
out home brew design that uses premixing, although one did appear 
in QST by a W1 ...   The TR4 series are premixed, right?   They're 
probably a good place to look.

2.  I need to see a Pullen mixer with circuit values and performance 
measurements before I'll be convinced that the thing has much to 
offer.   From my experience (which may possibly translate as "my 
inept attempts to build one") it appears to be a low level switching 
mixer -- that is, quite linear on the signal path for small signals.   If 
so, its use is restricted to no-RF gain designs.   

3.  I'm not clear on the value of re-converting the IF before the final 
detection.  Why shouldn't the BFO be on the IF?   With a high gain 
IF you have to 'live clean' anyway -- shield and decouple most 
carefully, so BFO pickup at the IF input isn't a problem; with a single 
IF stage the answer is the same but easier to attain.

4. Triode audio:  Nah ... use feedback.   And to make the best use of 
that, you need bunches of extra gain -- use pentodes.   Start with the 
RCA tube handbook hi fi amp designs.

5.  The idea of using plug-in premixer subassemblies for the various 
bands is interesting.   But I don't see any real problem with 
bandswitching this, and ganging the tuning of the output with that of 
the RF/Antenna coils -- it's just the regular superhet tuning problem 
except that instead of tuning the vfo you're tuning the output of the 
premixer.   Just need a variable cap, with a whole bunch of gangs 
and a single coil/fixed cap selected with the crystal for each band, 
since the higher bands (at least) will require overtone crystals.   
Complexity ... it's an issue, but going plug in just gives you a 
different kind of complication.  Or so it seems to me.

Walt
KJ4KV