[HBR] HBR2K Chapter 11 -- Sensitivity Training

[hbr@mailman.qth.net]

Previous chapters got it working but the sensitivity wasn't that hot.
Increasing the first oscillator injection to the limit of what the 
oscillator could deliver without drift improved things somewhat. 

In the process of changing that, I discovered that I had the grid 
connections reversed on the first mixer.   Doing it right rather than 
the alternative, improved sensitivity from 0.6 to 0.45 uv/10 db s/n on 
40 meters.

Then a lightbulb went on -- I had done alignment using the S-meter.   
What about our old friend the Miller effect?   Good sensitivity would 
require alignment with small signals.   And sure enough, alignment 
with less than a microvolt and the AGC disabled got the figure down 
to 0.25 uv/10 db s/n on 40.   That's much closer to what should be 
possible although the ultimate number will probably be somewhat 
compromised by the need for multiple front end traps at the 1st IF to 
get rid of HF broadcast station leakage.

('Miller effect' -- because the grid and plate are swinging in opposite 
directions the apparent grid to plate capacitance is multiplied by the 
voltage gain of the stage.   Since AGC varies the gain, it varies the 
apparent capacitance and a controlled stage can thus only be 
precisely aligned at one AGC level -- normally zero, if maximum 
sensitivity is wanted.)

I had suspected a bad RF amp plate trimmer on 80 and sure enough, 
it seemed to short out at certain places as I rotated it.  I had a spare 
in another parts set but what a miserable job changing the thing.  
These are conventional rotary ceramic trimmers but instead of being 
the size of a nickle as in most ham sets of the 50's, they're about the 
diameter of a lead pencil with 15 jammed in a space less than 1-1/2" 
square.   The back of the PC board is packed with fixed padder caps 
and coaxial cables and the trimmers are held by three tabs, two of 
them bent over before soldering.  

The next step is to equalize the injection across the bands or at 
least as nearly as possible.   Yaesu provided for that by putting 
loading resistors across the oscillator tank on the lower bands  
(Pierce circuit with a tank) but of course the loads are those 
appropriate for a bipolar transistor circuitry with very much lower 
impedences.   I think that will considerably improve things on 160 
and 80 meters.

The load resistors are in the mess on the underside of the trimmer 
boards.  But even with the need to replace a few parts, using the 
Yaesu assembly brings a job that would be all but impossible, down 
to one that requires a couple of hours of careful work.

I believe the drive is still low.   For the Pullen mixer, the drive should 
swing the tube to cut-off and with the variable-mu 6ES8, that's not 
easy to do.   I can get more by redoing the oscillator as a Miller.   
For equal crystal current a Miller circuit gives more output voltage 
than does a Pierce.

I need to study this; it may turn out that a good conventional triode 
like the 6DJ8 gives better sensitivity but the 6ES8 becomes the 
champion when overload/crossmodulation issues are considered.   

I believe there's another bad trimmer on one of the four 10 meter 
bands so that is something to look forward to.  I don't have all the 10 
meter crystals yet ... 

I had unstable regulated voltage because the neon lamp was 
unstable.   A new lamp fixed that.   Now waiting for the lamp to age, 
before making the final selection of the dropping resistor for the VFO. 
Yeh, another kind of instability -- neons increase maintaining voltage 
steadily for the first few hours of service.

The front panel plate is on order, I've figured out how to 'make over' 
the FT-101 bezel, a nameplate is made, and I'm working on the 
schematic and some construction notes -- 20 pages and counting.   
Somewhat hampered by the fact that a lot of the work was done over 
a year ago.    The worst is the VFO -- lots of critical mechanical 
detail that has faded from memory.   I will probably build another 
VFO before finishing the notes.

I have the first of four pages of schematic mostly drawn.   Found two 
more wiring errors in the process -- a missing bleeder on the negative 
supply and incomplete wiring of the switch section that should bias 
the carrier oscillator to cutoff when on AM.   I should have the 
schematic ready in a month or so.   

I believe that this set can be built with no more than a cheap DVM 
and a GDO and would take someone who has built simpler projects 
(certainly anyone who has done a Crosby HBR-xx) about six months 
of 'an hour or two, most evenings' to complete.   The parts issues are 
all managable -- all the fancy stuff comes in a 'parts only' FT-101 and 
nearly all of the rest can be obtained from OSE.   The construction 
notes will give a stage by stage procedure with tests at each stage; 
all the dumb things I did are getting written down.  The wiring is 
difficult in a few places but the worst of it will be better with some 
changes to the chassis layout to give more room in the tight places.

The front end is inescapably tight quarters but nearly all the wiring 
can be done before the bandswitch is installed -- the bandswitch, 
trimmers, and crystal socket can be gotten ready outside the set 
and then just dropped in, with about ten connections.  

Projects always seem to me to have their own styles.   For HBR2K  
the style is 'it wants to work but is sometimes hampered by my 
errors.'   No kidding, there have been almost no interesting problems 
getting it going, and the few that there were, were design problems 
that are over and done with.   

Walt Hutchens
KJ4KV