[HBR] GC-HBR Project -- Was 'Navy guy ...'

[N2EY at aol.com]

In a message dated 7/3/06 7:17:06 PM Eastern Daylight Time, 
waltah at earthlink.net writes:


> Jim, I, um, didn't REALLY think you snuck out the back way, but I 
> was starting to wonder if maybe we should get up a search party.   
> Here I'm throwing out the second or third long post about a paper 
> receiver and I have yet to hear anything about why it won't work!    
> LOL ...  

Two words: Field Day. 

FD #1 for me was 1967 IIRC and I haven't missed one since.

> 
> Of course the fact is that they don't all work -- the batting average 
> is around 50% I think -- not even that good if you count the paper 
> design.   So your comments are always appreciated.
> 

Thanks!  (where's my pencil?)

DEVIL'S ADVOCATE MODE = ON

> > Multiplying by 10 requires a quintupler and a doubler. That quintupler 
> isn't 
> > going to be very efficient. Worse, there's going to be the challenge of 
> > filtering out the 4th and 6th harmonics, which will not be very far away, 
> > percentagewise. And it's percentagewise that counts!
> > 
> > For example, if the 6000 xtal is quintupled, we have harmonics at 24,000 
> and 
> > 36,000 that we don't want, yet we want the 30,000 It's doable, but not 
> easy, 
> > and you'll need to switch the tuned circuits as well as the xtals.
> 
> This is of course the trade-off that one accepts in such a design.   
> A five-tube ham receiver has no RF stage, an R-390 cost what 
> would have been a couple years pay for most of us, the first few 
> decades of solid-state receivers were indeed compact, cool-
> running, and light-weight, but not too hot for hearing signals -- and 
> multiplying few-Mcs crystals up to synthesize a first converter 
> frequency will bring spurious signals that can minimized but 
> probably not completely avoided.
> 

Agreed - but for the effort the proposed rx would entail, paper is a lot more 
expendable than aluminum and solder. Of course since it's your aluminum and 
your solder....
> 
> 
> > > > Subtract 49.2 Mcs yielding 7.55-37.30 Mcs by 250 kcs steps.  
> > 
> > What bandswitch has 120 positions?
> 
> One with a whole lot of really teeny-tiny contacts?   But that's not 
> what I'm contemplating.  Let alone 120 tuned circuits for each 
> wafer.   The plan is a bandswitch with 9 positions, used in three 
> groups.   Each group of three selects the same front end tuning 
> range, roughly 1.75-5.5-15-30.5 Mcs.   The mixer input and 
> synthesizer coils are switched in those ranges and must be 
> manually tuned.   
> 

Interesting! But such a switch is going to have a *lot* of sections! And 
getting the isolation between sections isn't easy. Another lesson I learned the 
hard way from the Type 7

> For example, on the lowest band the 1st mixer is tuned 1.75-5.5 
> Mcs.  That's one knob.  Using the plan of a quintupler-doubler 
> might mean an untuned Pierce oscillator circuit (5.825-6.125 Mcs) 
> followed by the quintupler tuning 29.125-30.625 and doubler with 
> output at 58.250-61.250 Mcs.
> 

And tuned circuits all over the place! 

> The doubler drives a mixer that subtracts 50.7 Mcs (no longer 49.2) 
> so the synthesizer mixer output (to the receiver 1st mixer) must be 
> tuned 7.550-10.550.  
> 
> (Subtract the tunable IF of 5.550-5.800 Mcs from those frequencies 
> to get the coverage of each 250 kcs band.)   
> 
> The synthesizer thus has three tuned circuits that on the lowest 
> (1.75-5.50 Mcs) range cover 29.125-30.625, 58.250-61.250, and 
> 7.550-10.550 Mcs.   These can be ganged together, but NOT with 
> the front end tuning because the latter must also be retuned 
> WITHIN the 250 kcs band on the lowest few bands.   The 
> synthesizer is thus a second knob that must be set up on the 
> chosen 250 kcs band.  
> 
> (In a commercial grade version you'd use differential gearing and 
> perhaps a cam to combine the setting of the LO (the tuning within 
> each 250 kcs band) with the synthesizer tuning giving the mixer 
> tuning as output.  Or you could add a gang or two to the LO cap 
> and bandswitch the coupling to it to tweek a mixer that was mainly 
> tuned by the synthesizer.   But none of that would meet home 
> construction criteria.   I will use two knobs.)
> 

The problem I see is that the multiplier/mixer setup can generate a *lot* of 
spurious injection signals for the first mixer. Weak harmonics of the 
multiplied LO mix with the 50.7 and all sorts of hilarty ensues.

This may not seem a bother on paper, and after all it's not me building the 
thing.

But one of the big differences I find in receiver performance is what I call 
the Fatigue Factor. FF is influenced by a whole bunch of variables. Having a 
lot of spurious responses may give the rx a high FF. Of course FF is a purely 
subjective thing, and a receiver that is so good to me that I can't bear to 
turn it off may give you the screaming meemies in five minutes. 

> This lowest range covers 14 bands of 250 kcs each.   (1.75-2.00, 
> 2.00-2.25 ... 5.25-5.50)  Selection of the individual band is by 
> chosing the proper crystal.  The simplest way to do that is to put 
> one socket on the front panel and you just stick in the right crystal. 
> However since only three ranges are needed and a 9-position 
> bandswitch is available, I plan to allow selecting of either of two 
> internal crystals, plus a front panel socket, per range.   On the 
> lowest range the internal crystals would likely be used for coverage 
> 1.75-2.00 and 3.75-4.00 Mcs.   When another band is wanted, that 
> crystal goes in the external (front panel) socket.
> 

Big coffee can for the xtals. Also you have the problem that when you are 
tuning a range right next to or on top of the tunable IF, the feedthrough can be 
a real headache.

> Another wrinkle that affords some simplification is that the 
> synthesizer covers a much smaller percentage tuning range than 
> the receiver does -- just as a BC receiver front end tunes 550-1650 
> (3:1) but for a 455 kcs IF the LO covers 1005-2105 or just over 2:1. 
> I expect to combine the first two front end ranges 1.75-5.5-15 Mcs 
> into one synthesizer range.  
> 
> > Some ideas:
> > 1) The complex LO system can be big trouble. I learned the hard
> > way why premixer designs are so rare in ham gear: the LO has to be
> > *really* clean to avoid spurs. In a transmitter, if we have spurs
> > 60 dB down, it's no big deal. But in a receiver, if we are trying
> > to listen to a -130 dBm signal, and we have a spur that's 60 dB
> > down in the LO, and the spur is in the wrong place, we may hear a
> > -70 dBm signal that really isn't there. In the HF ham bands, -70
> > dBm signals aren't rare when the band is open and you have a
> > decent antenna. 
> 
> My thinking exactly.   That's why I went with double conversion.  
> It's just a whole lot simpler to break the spurious signal problem 
> into two chunks, producing first a crystal controlled signal that 
> covers a wide range and has fairly widely separated spurs -- the 
> strongest ones not closer than 5.825 Mcs -- and then separately a 
> tunable LO that covers a much narrower range at a much lower 
> frequency that (with care) can be kept out of the antenna circuit.

I'm afraid the synthesizer may have spurs all over the place. If the output 
of the x10 chain is not 100% pure, those spurs will mix with the high fixed LO 
and all sorts of fun will result.
> 
> For a ham-band only design, I'd shoot for a premixer setup.  It has 
> been done successfully by some good people -- Drake, right?   
> One could do far worse than copying the Drake numbers with 
> better mixers and tubes.   But for a general coverage setup ... no, I 
> don't think so.   

Actually the Drake setup works well for general coverage with one hole - 
right around the first IF.  Feedthrough!

The way Drake did it in the R4 & T4 was to use a first IF around 5700 kc and 
a tunable IF around 4400-4900 kc. (I forget the exact numbers, but I do recall 
that the heterodyne xtal needs to be 11.1 Mc higher than the band you want to 
cover). This scheme puts the  images and spurs way above both the IF and 
signal frequencies, at the cost of pretty high het xtal freqs on the higher bands.

The Hallicrafters SX-146/HT-46 pair did the 5 Mc VFO/9 Mc IF thing, as did 
TenTec. 

The Southgate Type 7 uses a premixer scheme devised to accomodate the 
available 1.4 Mc. filter. But it's not general coverage and does have some spurs. The 
first scheme tried for the Type 7 did not work well at all and had to be 
abandoned.

Almost all other ham setups used double conversion with tunable second IF.

For my money, the main reason to use a premixer setup is so that you can do 
single conversion and put the sharp IF filter knothole real close to the 
antenna, yet keep the tunable LO frequency low and avoid bandswitching it. If you 
are willing to do double-conversion-tunable-second IF, might as well go to 
simpler injection schemes.

> 
> > There's a circuit in a QST article about 1961 that allows
> > fundamental-mode FT-243s to operate in overtone mode. (It's in the
> > article by W1ICP that uses a single 6U8 and a 3500 kc xtal as a
> > converter for  40, 20 and 15 to 80 meters, without having to get a
> > costly xtal for 10.5 or 17.5 Mc. Being able to overtone those 
> > xtals might have some uses. 
> 
> The reason I'm not considering overtone operation is that although 
> I'm confident that at least most of these crystals could be operated 
> on overtones but the oscillator would have to be tuned and I don't 
> think that could be ganged with the multiplier/mixer tuning because 
> the relative setting is likely to vary from crystal to crystal.  
> Additionally, the 5th overtone might not be sufficiently stable on all 
> of them.  And the difference between the 5th harmonic and 5th 
> overtone will also vary from crystal to crystal, meaning a wider 
> calibration tuning range.
> 

I think all those problems can be solved - except the last. The 5th overtone 
is likely to be several kc off the 5th harmonic, and vary with the xtal..

> Ever notice the little gray or shiny metal spots on the corners of a 
> pressure-mounted crystal?  These crystals (FT-243's)  can't 
> achieve the Q of the wire mounted ones because there's a frictional 
> loss.  Add to that the reduction of Q by the fact of overtone 
> operation ...  Overtone operation of non-overtone crystals does 
> work but it's a per-crystal thing.   Plug and play requires a crystal 
> that was ground for overtone operation. 
> 

The circuit in the oscillator never failed me on any FT-243s. But if cal 
accuracy is an issue, it's not the best.


> Real crystals have an assortment of resonant frequencies 
> corresponding to different modes of vibration.   The Q and stability 
> of operation at the desired frequency depends not just on the mode 
> itself but also on energy coupling between the various modes -- all 
> the unwanted modes are just losses to the desired one.   The 
> design/production process aims to lower the Q of and coupling to, 
> unwanted modes.   Of course the desired modes are dramatically 
> different for overtone operation than for a fundamental.   
> 
> In the early WW-II time frame (BC-1335 crystals) this science was 
> in its infancy:  The state of the crystal art then was pretty much 
> that pressure mounted crystals in the few-Mcs range should be 
> slabs with the edges very slightly thinner than the center.  Wire 
> mounting was developed fairly early (the BC 603-604 tank FM set 
> crystals in the half Mcs range, beloved of post-war hams for 
> building crystal filters) but I believe overtone units did not enter 
> service until after the end of the war.  
> 
> The U.S. Navy learned its lesson the hard way.   The TDZ-RDZ 
> (transmitter/receiver) end of the war UHF sets used fundamental 
> crystals multiplied from the 5 Mcs range to 225-390 Mcs -- 
> catastrophic in task force operations due to all the spurious 
> frequencies.   This equipment had nearly a negative service 
> lifespan, something like 1947-49 which was a shame because the 
> 10-channel TDZ tx was a true thing of beauty -- six or eight Collins 
> Autotune units all going at once, tuning up all those multipliers.  I 
> was 9 when I first saw one, and the memory is with me still.
> 
> The successor TED-RED (single channel) used CR-24 'barrel' 
> crystals which were overtone units in the 20-30 Mcs range as I 
> recall.   These were far more practical radios and were still in 
> service when I was in the Navy, early 60's.  
> 
> The aircraft radios had it better -- probably because of space/weight 
> considerations, the first UHF sets used overtone crystals.  The 
> ARC-12 used CR-24 crystals in a turret inside an oven to control 
> the receiver and the receiver fed a discriminator driving a small 
> motor that frequency locked a TX master oscillator in the low VHF. 
> 
> You still sometimes see CR-24 crystals -- 9/16" cylinders with an 
> axial pin on each end -- at hamfests.  The most common frequency 
> is 30.375 Mcs, 1/8th of 243.0, the military UHF emergency 
> frequency.   All the early multichannel radios and many single 
> channel emergency sets used one of these
> 

All good info - but we don't have a barrel of CR-24s!

So let's back up and think about what we do have.

Seems to me that what you want is a general-coverage (160 through 10 with all 
the spaces in between) rx that will make best use of that barrel of FT243s in 
the 4-9 Mc range. 

The simplest way to do this I know of is to build a basic receiver and put a 
converter in front of it. Say we built a basic 1.75 to 2.25 receiver with IF 
in the 455 range (actually more like 465, to use some of those bazillion 
FT-241s). Such a receiver could use the tuning gang from a BC-454, so we'd have a 
slow tuning rate (10-12 kHz per turn or so), integral dial and ganged sections 
for tracking. 

Then there would be a conversion stage ahead, to be switched out on 160. To 
avoid feedthrough, we might compromise and give up the 2.25 to 3.0 range. We'd 
also compromise and have some of the bands tune backwards. While the 
backwards-tuning may seem annoying, it can be handled by a two-color dial or even a 
movable mask like W2LYH did. Sideband inversion may be a little more annoying but 
a good variable BFO will fix it.

Doing some quick back-of-the-enveloping, a het scheme starts to emerge:

3.0 to 3.5 would require an xtal at 5.25 (subtract)
3.5 to 4.0 would require an xtal at 5.75 (subtract)
4.0 to 4.5 would require an xtal at 6.25 (subtract)
4.5 to 5.0 would require an xtal at 6.75 (subtract)
5.0 to 5.5 would require an xtal at 7.25 (subtract)
5.5 to 6.0 would require an xtal at 7.75 (subtract)
6.0 to 6.5 would require an xtal at 8.25 (subtract)
6.5 to 7.0 would require an xtal at 8.75 (subtract)
7.0 to 7.5 would require an xtal at 5.25 (add)
7.5 to 8.0 would require an xtal at 5.75 (add)
8.0 to 8.5 would require an xtal at 6.25 (add)
8.5 to 9.0 would require an xtal at 6.75 (add)
9.0 to 9.5 would require an xtal at 7.25 (add)
9.5 to 10.0 would require an xtal at 7.75 (add)
10.0 to 10.5 would require an xtal at 8.25 (add)
10.5 to 11.0 would require an xtal at 8.75 (add)
11.0 to 11.5 would require an xtal at 6.625, doubling to 13.25 (subtract)
11.5 to 12.0 would require an xtal at 6.875, doubling to 13.75 (subtract)
12.0 to 12.5 would require an xtal at 7.125, doubling to 14.25 (subtract)
12.5 to 13.0 would require an xtal at 7.375, doubling to 14.75 (subtract)

(rest of table left as an exercise for the reader)

You can see where this is going! There are probably somewhat better schemes 
using fewer different xtals, but the concept is the same.

The xtal oscillator can be the familiar grid-plate circuit used in 
transmitters, with a smaller tube than the ubiquitous 6AG7 (6AH6 would be a good 
choice). The output tank would be bandswitched and tunable as described in your rx. 
If your xtal set reaches 8950, doubling will get you to about 21 Mc and 
tripling to 28 Mc or so. 

In fact, if one were really careful, it might be possible to cover all the 
segments with 12 crystals or less. In that case, a standard 12 position switch 
could be used to select the xtals - no front panel socket at all. The het osc 
output tuned circuit (with peaking knob) could be a double-tuned affair going 
from 5.25 to 27.75 Mc - that would take three positions at most. 

So you'd have three bandswitches - receiver input, xtal select, het osc 
tuning - but they'd be relatively simple switches. Two peaking controls - Input 
Tuning and Het Osc Tuning. 

A chart would tell you that to receive, say, 17.5 to 18.0, you'd choose 
(values are for illustration only) 

Input Band 2
Input Tuning 57
Xtal D
Het Osc Band 1
Het Osc Tuning 44
Red scale on the dial for tuning.

Once selected, only Input Tuning would need to be peaked up as you tune 
across the band.  

Such a receiver could be very stable because the tunable LO would be below 3 
Mc. The fixed IF could be 455 or thereabouts, depending on whether you used 
FT-241s or something else. Image rejection at the top of the range might not be 
perfect with an IF down around 2 Mc, but it would not be any worse than the 
classic HBRs with their 1.6 Mc first IF. 

Most of all, the rx could be built in sections: 

Audio
455 filter/IF/detector/BFO
Tunable IF/LO
Downconverter
Heterodyne oscillator

--

Of course it's one thing for me to do a paper design and another to actually 
build the thing!

73 de Jim, N2EY