[HBR] HBR -- Part 8

[Walt Hutchens]

> There's a trace of hum (at full volume) on the (pure) signal from
> the GDO so that will have to be looked into. Most likely,
> heater-cathode leakage, probably in the plate detector stage. Second
> choice, inadequate filtering of some high voltage.

The first guess was the winner: Heater-cathode leakage in the plate
detector stage.  Remember when the best high-fi amps applied a few
volts positive bias to the heaters?  That's what they were fixing.

The 50-watt amp in the last RCA Tube Manual has this circuit.  You can
do the same thing (if no simpler approach works) by returning the
cathode to a negative supply.  This voltage must be a few volts
greater than the peak filament voltage.

The cathode sleeve acts as a plate for a diode having the filament as
a cathode. When the heater swings negative relative to the sleeve some
electrons reach the cathode and flow in the cathode circuit. This
effect is unimportant in most circuits (any time the cathode is
grounded for signal AND 60 cps frequencies) but in high gain circuits
with a 'hot' cathode, it must be reckoned with.

The plate detector cathode is where the BFO is injected for SSB, but
I've found that I can ground it for 60 CPS through a small choke and
0.1 mfd cap; that eliminates the hum on a strong carrier.  Also it is 
grounded when on AM anyway.  

> One definite problem is that there's an offset of about -0.7V on the
> AGC line, evidently coming from the vacuum diode I used in this
> circuit. "Shucks, I should have thought of that." There are some
> options for dealing with this; I don't know which I'll try first.

The reason for this is that even with no voltage applied to the diode,
there are some electrons striking the plate, hence some negative
current flowing in the external circuit having a resistance of a bit
over a megohm.  One microamp would give one volt ...

This turns out to have a very neat solution: Reduce the cathode
resistors so that the cathode bias is about +0.3 volts, giving a total
grid bias with zero signal of about 1 volt, as desired for the 19JN8
pentode.

I've seen a commercial-grade receiver that had no cathode resistors at
all: they used the AGC to do the whole job. No protection for the
tubes that way, though, if the AGC fails.

> The next construction step will be the crystal filter, 1st IFT, and
> mixer circuit.

Starting with 100 turns of #40 on a 1/4" diameter slug-tuned form.
Ugh. However, this time I didn't break it the first couple of times.

Testing of the two IF stages showed three more problems:

1. With a modulated signal, the audio amp was driven into distortion
before the AGC kicked in. I ran into the same issue with the earlier
receivers using this (plate detector) AGC circuit and the solution is
to tap the last IFT so that the voltage going to the AGC is
considerably higher than that going to the audio detector. I switched
from two 300 mmf caps in series across the IFT output winding, to a
200/620 mmf combination giving about a 3:1 ratio and that makes things
a LOT better.

2. When the AGC bias control (RF/IF gain control) is reduced, you can
hear hum, which goes away as you continue to reduce the gain.  That
seems likely to be a bypassing problem around the diode; I haven't
looked into that yet.

3. The 3rd IFT didn't peak up correctly on the AGC voltage when I put
a signal through from the 1st IFT. This turned out to be a bad
coupling cap from the 3rd IFT to the grid of the AGC detector.

Of course 'turned out to be ...' was after substituting tubes and
removing and disassembling the IFT, looking for a short or something.
There's probably a rule that every worthwhile project will include at
least one bad part.

That's a corollary to the rule that some items in your junkbox are
junk.

The gain of the IF/detector chain may be a bit low. I'll worry about
that later on, if necessary. There's plenty more AF gain available,
and since that's set by a feedback resistor, it's very easy to change
if needed.  And there are some things to check in the IF stages.  The
voltages are in the ballpark, but the plate current of the tubes is
low; I think the screen voltage is a few volts below what it should
be.  Also the tubes might be sub-par; while they're claimed to be new,
I haven't checked them.

I probably don't even have data for 19JN8s for my TV-7. However the
data can be figured out by looking for single section tubes equivalent
to the respective halves.

Testing is easier, now that the S-meter is working.

The half-lattice crystal filter is in and working.  Using crystals
that are 1.5 kcs apart in frequency, the nose is reasonably flat and
about 2.2 kcs wide.

Next steps are wiring the mixer and winding the front end coils.  I'm
thinking that I will use the W6TC method of bandspreading these,
namely, tapping the tuning cap down on the coil.  I have a cap meeting
the HBR specs (delta C = 15 mmf) so there's no reason that I can't
just wind the front end coils using the original HBR specs.

Walt
KJ4KV