[HBR] Another Project -- 1MHBT -- Part 3

[waltah at earthlink.net]

First, some pictures of the 1Month HBR -- this is the receiver 
project that's about to get a companion transmitter.   

http://www.timbreblue.com/radio/walt3_images/3.htm
http://www.timbreblue.com/radio/walt3_images/4.htm
http://www.timbreblue.com/radio/walt3_images/5.htm

I don't seem to have pictures of the chassis out of the box.  I'll do 
that in the next couple of days.

Turning to the 1MHBT -- can it possibly take over a month, once I 
start?   

(Yes, I hear the laughter out there.   But I'll give it shot, anyway.)

Another data point on suppression of the unwanted sideband when 
using 85kcs command set IFTs:  According to the ARC-5 
maintenance manual, the selectivity of the 0.19-0.52 Mcs ARC-5 
command receivers is: 

Bandwidth    db down

1.1. kcs       6 db

2.2  kcs      20 db

3.3  kcs      40 db

4.5  kcs      60 db 

These measurements are taken at 190kcs, but the front end coils 
aren't high enough Q to have a significant effect.  

If in the transmitter design the audio frequency at center band is 
1.5 kcs, then the 6 db bandwidth covers 950 to 2050 cps, the 
carrier will be down about 30 db, and 500 cps on the opposite 
sideband would be 50-some db down.  With another 20-30 db 
carrier rejection from the balanced modulator (easily obtained), 
carrier suppression should be ample.   

That's with three transformers; with the planned four, the db down 
at each bandwidth should be 4/3, ignoring leakage issues.   Of 
course you can't ignore leakage and both the layout and the 
shielding will have to be planned to keep it to a minimum.   

It won't sound like a modern design built around a crystal filter with 
a flat top: "Audio's a bit 'peaky,' OM."   Well, it's not a modern 
design ...

Depending on the ARC-5 model these selectivity figures are 
measured either with the rods in all three transformers pushed 
down (so they are overcoupled) or with one down and two up.  I use 
the more selective transformers and run the 1MHBR set with all the 
rods pulled up.  Audio is certainly 'peaky' but plenty 
understandable.

A design issue:  If series filaments are used including the plug-in 
PA section I've been considering, then changing bands requires 
turning off all the filaments -- rather hard on VFO stability.   
Alternatives would be separating the PA filaments from the others, 
operating all filaments in parallel, separating the VFO filament, and 
finally, taking the PA out of the plug in unit.    

And *that* issue was the start of several days of chewing on the 
basic idea.   The harder I looked, the more problems.  In addition to 
the filament issues there wasn't enough room for a plug in PA unit 
if the VFO dial was to be anywhere near the center of the panel 
(matching the 1MHBR), there were heat flow issues, and there 
were shielding and chassis current considerations.  Additionally I 
wanted to use somewhat larger tubes than 22JF6's if possible.

The space problem is fundamental and difficult.  Usually the PA is 
placed at the rear of the chassis with long shafts to the front and 
the driver and earlier exciter stages wrapping around the front and 
turning back toward the PA.   However anything that plugs in 
through the front panel has to be in the front.  If using a VFO tuning 
cap with an integral dial as for the ARC-5 receiver tuning cap I'm 
planning on, then the VFO must also be in the front.  

A push-pull PA is appealing -- you get twice the horsepower of a 
single tube without the balancing that is becoming ever more 
difficult as tubes become as-is, where-is -- and you get some 
cancellation of distortion products.   (Two tubes in parallel are 
worse than one unless perfectly matched; two in push-pull can be 
better than either.)  But push-pull makes symmetry demands on 
the relative locations of coils, tubes, and tuning caps and some of 
those make the space problem harder.

Taking the PA tubes and tuning caps out of the plug-in units 
seems the best answer and of course saves some work and some 
money.  I *think* I've got a layout that works, with the PA tubes (2" 
width required) all the way to the left on the chassis, a 4" high x 2" 
wide x 6" deep chassis holding the grid and plate coils (about 2-
1/4" width required) sliding in just to the right of the tubes and the 4-
1/2" tuning cap barely to the right of chassis center.  With a 14" 
wide panel I can shift the 12" x 10" x 3" chassis left by up to 1/2" 
relative to the panel to keep the dial centered in the panel.

The plug in coils make this a mechanical mess just as was true for 
the receiver project.  The tuning unit will go through a 2" wide x 4" 
high hole (+ clearances) in the panel into a frame made of 
aluminum angle stock ... a couple of pieces to be cut, folded, and 
screwed together.  Contact buttons on the side next to the tubes 
for the plate connections and on the bottom for the grids; finger 
contacts matching each, probably also a plug on the rear for the 
center taps and the input and output connections.   There may 
have to be a shield partition in each coil unit between the plate and 
grid coils -- a toroid should work for the grids and that's the only 
reason for the 'may.'    However, with separate coils on each band, 
neutralization should be a cinch!

PA tubes will be submounted about 1-1/2" to gain clearance above 
for the plate tuning cap above the tubes.  Small fan pushes air into 
the chassis at the rear; it exits through the clearance holes around 
the PA tubes and up through the VFO tube clearance hole -- that 
socket will be mounted just enough above the chassis hole (about 
1/4") to allow air to come up around the pins, minimizing the heat 
piped down the wiring to the tank circuit.  Probably also small 
holes in the chassis to circulate air up behind the panel to 
minimize heat transfer to the VFO tuning cap.

Some of the postwar Collins who-cares-about-the-cost airborne 
radios have a pressurized chassis with calibrated ventilation holes 
all over the place, some of them as small as 1/8".   There's no 
need to do anything that elaborate in a ham set (cooling a 50-tube 
radio that operates at 40,000 feet is a harder problem) but thinking 
about heat flows as the design is worked out can pay large 
dividends in stability.  The 1MHBR is decently stable only because 
of it -- drift is under 1 kcs on all bands including 20 (no coil for 10, 
yet) despite an old-style (oscillator spread around) design and 
rather casual temp compensation because ventilation limits the coil 
temp rise to about 10 degrees F.  

Assuming all those problems get solved, the tubes and tuning caps 
will be on the chassis with just the PA plate and grid coils in the 
plug in tuning unit, meaning that a series filament design will be 
possible.  

There are at least six choices for the PA but the front runner just 
now is the 42KN6 with a seated height of 4", a 12 pin base, a peak 
current at the 0 grid volts 'knee' of around 800 ma, a rated plate 
dissipation of 30 watts. and, last but not least, an Antique 
Electronic Supply price of $9.30 -- about the minimum for anything 
in this power range.  

A pair should hit about 75+ watts pep at 480 plate volts, over 100 
at 600 or so.   Transformerless plate supplies get pretty bulky if 
you go above a tripler (both the number and the size of the largest 
of the caps are in proportion to the multiplication, thus size goes 
up as the square of the multiplication), so the lower voltage may be 
more practical.   Or maybe not -- unless lots of RF clipping 
(speech processing) is used, a linear amp for voice signals doesn't 
need a high average current capability.   

A 6U8 audio amp, three 6JH8 beam tube balanced mixers, and four 
4EH7's (two 85 kcs IF's, a driver and probably a crystal oscillator) --
that's eight tubes totalling 40 volts.  Adding the two 42KN6's 
makes the total filament voltage 124 volts.  If more filament voltage 
is needed for a presently unplanned stage the 31JS6 would be a 
near drop-in alternate and the 36MC6 and 31LZ6 are possiblities  
(but with 9-pin bases and higher prices). 

Too bad the sheet beam tubes died so early; there don't seem to 
have been any of them made with 450 ma filaments.   Hard to 
believe that -- does anyone know of one?   Hummm ... might 
parallel a 12AX7 audio amp (replacing the 6U8) with two of the 
6JH8s to save some filament voltage.   Or maybe a 19JN8 (triode-
pentode), paralleling all three 6JH8's?   That would make the 
filaments total to 118.   

Walt 
KJ4KV