[HBR] HBR-11/2000. Comments?

[N2EY at aol.com]

In a message dated 7/13/06 10:30:53 PM Eastern Daylight Time, 
waltah at earthlink.net writes:


> I'm going to piggyback on Jim's comments on Barney's HBR 
> project.   Mileage may further vary ...

This is an old posting, but I'm not sure if I replied to it or not. I 
apologize for the delay...
> 
> >> After spending the last couple years collecting parts (including 
> >> the Eddystone 898 dial) I have decided to begin construction. 
> 
> > Congratulations!
> 
> YEP.   Just finding all the parts is an accomplishment.

It's often the main limitation. Long ago, I adopted the philosophy of 
designing around available parts because some once-popular parts had become 
unobtanium. Or appeared to be.
> 
> >> 2) 6ES8 Pullen mixer (ala 75A4 mod on the Web)
> 
> > Replacing the RF stage/pentode mixer with the Pullen may result in
> > loss of overall gain. I don't know exact numbers but you may want
> > to do some calculations and measurements before putting holes in
> > the aluminum. 
> 
> Overall gain, however, is not going to be a problem in this design.  
> 6EH7's have plenty to overcome the reduction in front end gain.   

Maybe. OTOH, 6EH7s at full gain can be hard to stabilize. And given the 
complexity and parts/time investment of the project, I would still highly recommend 
a few calculations and measurements before doing serious building.
> 
> A bigger issue is that Pullen mixer.   It's my impression that this is 
> an excellent weak signal mixer but not a remarkable performer with 
> strong signals.   (Jim, please correct if that's not so?) 

Not according to the literature.

I don't have any substantial experience with Pullen mixers, nor do I know of 
any published numbers for that design. 

There are excerpts from Keats Pullen's book online (see the glowbugs website) 
which detail the entire design process for the mixer, working from tube data. 


However, the big selling points for the Pullen were that it was reportedly 
both low-noise and good on big-signal performance, yet could be implemented with 
fairly common tubes and conventional single-ended high-Z "tube type" tuned 
circuits. And a wide variety of tubes could be used. 

Perhaps the best selling point of the Pullen mixer was that it was easily 
adapted to existing designs because there are so many dual triodes available. For 
example, in W2CVZ's mods to the 75A-4, he replaced the original 6BA7 first 
mixer with a 6ES8 Pullen, reportedly getting much lower noise and better 
big-signal performance without major surgery. And the mod is 100% reversible. Many 
existing receivers could benefit from similar no-holes modifications. (I've been 
thinking of seeing what a 6SL7 Pullen could do for my BC-342-N)

Whether a good Pullen is "better" than, say, a 7360 balanced beam-deflection 
mixer as used in the SS-1R I simply don't know.  I do know it's a lot easier 
to implement.

  Except 
> 
> possibly on 10M, low noise/weak signal performance (same thing) 
> is not the criterion for HF receiver design.   Atmospheric noise will 
> overcome any reasonable front end noise on the lower bands -- 
> only a 6K8 or 6BE6 will be limited by tube noise.  

I must add a big caveat here.

I say that in *most* HF receiver applications with a good antenna, there's a 
practical limit to noise figure/sensitivity, because the noise coming down the 
lead-in will totally override the receiver noise. However, whether a 
particular rx has enough low-noise performance depends on the band in question, the 
antenna in use, and the level of noise at the site. There's also the question of 
how the receiver is used.

If you have a big antenna and a noisy site, and operate mostly below 10 MHz, 
sensitivity isn't going to be a problem. 15 or even 20 meters with a small 
antenna in a quiet site is a whole different story.

 I'd use a more 
> 
> conventional triode mixer circuit, operatin g the mixing tube at 
> higher voltages to improve dynamic range -- in particular, 
> 'interference free dynamic range' -- the range of signals over which 
> strong signals do not mix to produce spurious signals.  
> 
> The 'strong signal' dual triode mixer using a 12AT7 with the 
> antenna signal on one grid, the oscillator on the other, cathodes 
> joined and to ground through 470 ohms, and IF take from paralleled 
> plates is an excellent first mixer for an HF set if used with high Q 
> coils -- you need those anyway.   The 6J6 also works well, a 6DJ8 
> would probably give you more gain.   Something around a couple 
> volts p-p injection.
> 

The nice thing about all this is that both that circuit and the Pullen could 
be tried without major changes to the design.



> Or you could try a push-pull triode mixer -- signal in push-pull on 
> the grids, LO in to the cathodes in parallel, IF signal out in push-
> pull from the plates.   I realize the issue with your plans for 
> bandswitching but that has its own issues.   
> 

Balanced designs (push-pull) really do complicate things, though. That's one 
reason the balanced beam-deflection mixers never caught on beyond the SS-1R.


> >> 3) 6DJ8 1st local oscillator, Vackar design
> 
> > I suggest two references before you finalize that idea:
> 
> > A two-part QST article called "VFO Stability - Recap and
> > Postscript" about 1966. Although a transmit VFO is described, the
> > issues discussed are right on for receivers. 
> 
> > A QST article called "Stability with Simplicity" by W2YM, about
> > 1961, and a following article "An Easy To Build VFO" by W1ICP. A
> > simple but very stable VFO circuit is described. I have used it
> > with excellent results. 
> 
> The Vackar is a good oscillator.   There is no magic circuit that's 
> more stable than the others -- the two issues are minimizing 
> coupling to the tube, which calls for a high gain tube, and achieving 
> a stable tank circuit.   The Vackar circuit gives you the easiest 
> control of coupling.  The 6DJ8 has about as much gain as you're 
> going to find.   

I agree there's no magic circuit, but the articles cited detail the problems 
involved. Many Vackar designs don't ground the tuning cap rotor, which makes 
things more complex.

I have often wondered why so many complex amateur receivers used very simple 
LO designs with no buffer stages. I suspect that stability issues in some 
designs are really the result of varying load on the LO.

> 
> > 4) Pair of Miller 1731 link coupled coils (diode DBM in between for noise 
> > blanking)
> 
> An interesting idea.  Though I hate to see solid state creep into an 
> otherwise perfectly sound design -- HI.
> 
> >>  5) 6AR8 sheet beam single balanced mixer (like the 7360)
> 
> > The 6AR8 is reportedly microphonic. That may or may not be a
> > problem. I don't recall any info about the 6AR8 as a receiving
> > mixer. 
> 
> The 6AR8 was the first effort at a controlled-cost beam deflection 
> tube.   The 6JH8 and 6ME8 are better tubes -- more stable and 
> less microphonic.   Run the deflection plates a few volts negative if 
> you want the greatest sensitivity for oscillator injection.
> 
> However I'm not sure a beam deflection mixer is worth the effort (or 
> even a good idea) for a second mixer.   You don't need anything 
> like that much gain and the sensitivity has been set in the first 
> mixer.    
> 

Agreed - mostly. The main reason to use a beam-deflection mixer is to get the 
combination of low noise and big-signal performance.

However, in a receiver with no RF stage, the second mixer noise may still be 
significant. This is where calculations are important - how much gain will 
there be from the antenna jack to the grid of the second mixer?

> Most hams figure the more gain the better.   The truth is closer to 
> the reverse: Use just enough gain to get over the noise of the 
> following stage.   
> 

But by how much? 



> >> 8) 6EH7 IF amplifier (lots of shielding)
> 
> > The 6EH7 is a really "hot" tube - lots of gain and low noise. The
> > Southgate Type 7 uses two of them. Be warned that they can be
> > difficult to stabilize, and at 100 kc you may be better off with a
> > 6BA6. If you allow room for a 9 pin socket, but use a 7 pin in the
> > space, you can convert later. 
> 
> One word -- shielding.  Oh, and good grounding and bypassing.   
> Just do everything right from the start and you'll have no trouble.  
> You do need a 100 ohm resistor right at the grid (1/8" lead) to kill 
> VHF parasitics -- this is needed for any of the modern high-gain 
> wideband tubes.   
> 

Maybe. Two stages of 6EH7 is a lot of gain! With only two stages, both of 
them mixers, before that IF, that gain may be needed.

> This is a superior IF tube, not just because of high gain/low noise 
> but more importantly for controlled intermodulation distortion.  I 
> wouldn't use anything else without a good reason.
> 
> Feeding a 6BN6 these tubes will only be loafing -- they can handle 
> tens of volts without distortion but you're limited to under one volt at 
> the output of the 2nd IF.
> 
> >> 12) 6BN6 product detector
> 
> > Shock mount it - that tube is microphonic too.
> 
> A fine choice for a product detector.   And simpler than using a 
> beam tube.
> 

Agreed. Also a pretty common tube. 

> > 14) 6xxx infinite impedance detector for AM
> 
> Consider a plate detector -- except for large signal handling ability 
> they have the same advantages as infinite impedance and give you 
> a better match to the gain of a 6BN6 as a product detector.   
> 
> You'll notice that receivers that use the lineup you propose have an 
> extra audio stage following the inf. impedance detector.   
> 
> Unless you copy an existing circuit exactly you'll have to tweek the 
> plate detector to get really good linearity.  At 100 Kcs you'll need 
> the 2.5 mH choke that most circuits show in the plate circuit.
> 
> You have to keep the signals down at the IF output anyhow 
> because the 6BN6 will take less than a volt before limiting sets in.
> 

That limiting can be a good thing in some cases. But you don't want the 
detector to limit on wanted signals!

> >> 15) AGC will be solid state - there is a lot of potential for creative 
> >>  design here
> 
> > Yes - but with all those bottles, why not hollow state?
> 
> Yeah ... why give up while you're ahead?   
> 
> It's hard to beat the AGC circuit of the Tempo ONE.  They use a 
> plate detector with the cathode biased negative so the plate runs at 
> ground level; signals pull the plate negative to generate the AGC 
> signal.  You do have to have a bias supply of negative 50-100 volts. 
> 

There's a number of easy ways to get a small negative bias voltage in a 
receiver. One way is a (cough) solid-state voltage multiplier off the heater 
circuit. Another is a half-wave off one side of the HV transformer, with suitable 
high resistance RC filter.

A third way, particularly if you have some extra heater power, is to run a 
small filament transformer backwards to get isolated 120 volts.

> IF-derived AGC for SSB gives you the challenge of keeping the 
> BFO out of the AGC.   If you don't want that challenge, then you 
> have to go audio-derived which has a different set of issues.
> 

Personally I don't see why AGC is such a "must have" to so many hams. 

> >> 16) 5879 low noise pentode 1st AF with lots of negative feedback
> 
> > 5879 is an audiophile tube - may be hard to get at reasonable
> > price. You don't need  or want a lot of audio gain. 
> 
> Any of the usual pentodes will do fine in this socket.   A com 
> receiver does not require the audio dynamic range of a 100 watt 
> high-fi amp (you don't need to hear the 3rd violin turning pages) and 
> you have that AGC loop getting rid of dynamic range, so extreme 
> low noise in the input stage is not worth the trouble.   
> 

The signal level at the start of the audio chain will be? I would guess over 
100 millivolts, in which case a much more common tube would do the job easily. 


> However you'll be surprised what using a low-distortion high-fi 
> circuit for your audio amp will do for voice clarity.  If everything up 
> to there is working correctly you WILL hear the difference.   That 
> means using feedback with enough loop gain that the feedback 
> eliminates distortion in your amplifier.   The high-fi amp circuits in 
> the RCA tube manual are a good starting point.  You must cut off 
> the amp frequency response much lower than for high-fi because 
> you won't have a transformer that will handle much over a few Kcs 
> without enough phase shift to lead to instability -- but for 
> communications use you don't want lots of highs anyway.
> 
> Use a two stage amp with feedback from the voice coil winding to 
> the cathode of the 1st stage.   Better yet, use push-pull output 
> stage with a phase inverter and do the same.  
> 

I found a noticeable improvement from all-triode audio. Pushpull triode 
output has much lower distortion (without feedback) than a single-ended pentode.


> >> 17) 6CW5 power amplifier (lots of audio power is important)
> 
> > Good. You can always bias the tube more to cut down on output, but
> > inadequate audio is hard to fix. 
> 
> The 6CW5 is overkill -- with a pentode 1st audio you've got plenty 
> of loop gain even with a triode output stage and something like a 
> 12BH7 is plenty.   Push pull lets you go class AB so you save 
> power while reducing the distortion so your feedback loop has less 
> work to do.   
> 
> Exception: If you want to drive true high-fi speakers the efficiency is 
> so low that you'll want more audio power.  In that case a pair of 
> 6CW5's wouldn't be out of line.   
> 
> Standard (modern) automotive 3" or 4" speakers are a good 
> compromise between 'fi' and efficiency.   The One Month HBR uses 
> a pair of 6AQ5's running at 150 volts and a 3" speaker and you can 
> listen to that one in the front yard when it's in the back.  And that's 
> with the volume just past the halfway point.   There's LOTS more 
> available, though it might be a bit much for the speaker.   
> 
> The R8040A uses half of a 12AU7 as an audio output stage driving 
> a 3" communications speaker.   Volume is ample for all but the 
> noisiest locations.   
> 

Some folks really like a lot of audio. Room acoustics can be an issue, too. I 
like big speakers because they are more efficient and usually roll off the 
highs naturally.

> >> 18) 190V 200mA power supply, the iron is Triad hermetic 
> >> isolation transformer  
> 
> Keeping the plate voltage down is an excellent idea.   The latest 
> tubes don't need it (in fact ratings may be exceeded) and lower 
> voltage means less dissipation with all the problems that go with 
> that.  
> 
> > Voltage regulation for the oscillators?
> 
> Probably a good idea for the 1st oscillator since it operates well 
> into the HF.   The crystal oscillators certainly shouldn't need it.
> 

Doesn't hurt, though.

> > Other comments:
> 
> > Reed-relay-switched toroids may not be the best for HF oscillator 
> > stability. The original HBR design coils and sockets are a big 
> > part of the stability. 
> 
> Bandswitching is a challenge.   It can be done, but I certainly 
> wouldn't start there unless you have some receiver experience.      
> 

I think one reason W6TC used plug-in coils was that he simply could not find 
a bandswitch that would do the job as well. The RF coils aren't hypercritical 
but the tunable LO is. 

Note that about 1960 or so, practically all 'serious' ham receiver designs 
got away from bandswitched LOs and went to other methods (such as tunable second 
IF) of getting multiple bands. Stability and dial calibration were the 
issues. 

73 de Jim, N2EY