[HBR] Mixer questions and other tech questions..

[Walt Hutchens]

Marvin said:
> These discussions are the most fun I've had on the 'Net in years. Keep those
> ideas coming.
> 
> While I love the discussion about tube rolling, readers should note
> that the receivers we're talking about only loosely resemble the
> HBRs. The HBRs have 2 "magic" features, as far as I'm concerned.
> Firstly, they are completely repeatable. If you follow the design,
> it's hard to screw up. Secondly, is the rather "elegant" front-end.
> Both these features disapear in the reciever design we're discussing
> here.

Agreed on all!

>> That means that the 1st mixer almost has to be singled ended, since a
>> balanced oscillator output over this wide range is a very difficult
>> job.  On 10, for example, stray capacitances will greatly affect the
>> amplitude and phase of the mixer drive. 
> 
> The LO can still be single-ended, just follow it up with a
> long-tailed pair buffer stage. That gets you the balanced output but
> allows you to use the conventional single-ended oscillators that we
> all know and love, and you stand a chance of developing enough drive
> for the deflection plates of the mixer to boot.

Except on the lowest bands any kind of phase splitter is going to get
you into both amplitude equality and phase angle issues unless the
push-pull output is tuned and tracks the oscillator. That's a lot of
complication.

>> Trying to make this work with conventional plug in coils would be a
>> real challenge. You COULD try doing it by building your own
>> balanced plug-in coils like the ones used in transmitter push-pull
>> stages.
> 
> Trying to build a balanced LO sounds really fiddly to me. Why
> bother? I don't think you'd ever get it right.

Actually, it's trivial for a single band because the tuned circuit
forces the proper 180 degree phase and equal amplitude relationship.
With plug in coils I don't think it would be difficult for an all-band
set, though you'd need quite a plug.  The tuned circuit has seven
parts hooked in a circle; one point is grounded and midpoint of the
coil is the B+ connection.  I think an octal tube base would do it.

>> Maybe use miniductor and tube bases?

> Or torroids?

Toroids in the tank of a vacuum tube oscillator have the disadvantage of
introducing more 'stuff' with a temperature coefficient into your
tuned circuit. In a solid state set, of course, that's much less an
issue.

>> These receivers have an RF amplifier.  They don't need a high gain
>> mixer.  If you reduce the oscillator drive to the point where the gain
>> with a beam mixer is no more than needed, then you've got pretty much
>> what you'd have with the original HBR design.
> 
> On the other hand, if you use a high-gain beam mixer, you can eliminate the 
> RF amp stage, or swap it for a low-gain G-G RF stage with barely enough 
> gain to cover the losses in the preselector.

You don't need the gain of a beam mixer. You can build a pretty decent
receiver for all the lower bands using a triode mixer and no RF stage.
Only on 10, maybe 20, might this not be the case.

The G2DAF Mk II receiver (RSGB Handbook 4th edition I think) has an RF
stage with less than unity gain; it's there only to allow AGC control.
This receiver has a very good (quiet, high gain) push-pull triode
mixer.

>> On the lower bands -- say through 20 meters -- the biggest issue in HF
>> receiver front ends isn't gain or noise, anyway, but performance when
>> there are multiple very strong signals within the passband of the RF
>> tuned circuits.   If any stage is driven into non-linearity even by a
>> signal you're not actually hearing (because it is eliminated by the
>> IFs) the result will be distortion of the signal you're listening to.
>> This is 'crossmodulation.'
> 
> See the receiver designed by Doug Blakeslee found in a 1972 QST. He 
> chose a 7044 dual triode, G-G, with both sections in parrallel for 9 watts of 
> plate dissipation.

That's a good triode RF amp. Grounded grid has the advantage of
providing some negative feedback which improves linearity. However,
the noise advantage of a triode RF stage (over a good pentode) isn't
significant on the lower bands when front end noise is tens of dB
below atmospheric noise and band 'trash' anyway.

About the only time that front end noise becomes an issue on the lower
bands is if you have a pentagrid tube as a first mixer. Then you MUST
use an RF stage with substantial gain. The solution is to take all
your 6SA7's and 6BE6's to the back yard and throw them as far as you
can.

Vacuum tubes might come back, but not the pentagrid mixer tube for HF
receiver service. It belongs in 'All American 5' receivers, where
signals are in the tens of thousands of microvolts and you need a
converter tube that can be AGC'd without pulling the oscillator ...
and nowhere else.

> ---snipped stuff about using a 6EH7 in the RF stage---
>  
>> This is no harder to get working than the original HBR circuit,
>> although the greater care you use in wiring the 6EH7 RF stage, the
>> more of the potential gain you'll be able to use on 20 and 10 meters
>> where high gain is sometimes useful.
> 
> Do you think you could get a 6EH7 to work as an RF amp in the HBR with 
> plug-in coils? I don't. You can't make the wires short enough.

Sure you can.  At the same gain level, you'll have exactly the same
amount of trouble getting the wires short enough as you would with a
6SK7 -- or a bit less, due to the superior basing.

It is true that at the maximum gain the tube is capable of, the 6EH7
is going to be more difficult to tame than more conventional tubes,
because it is capable of much greater gain.  6SK7 Gm = 1700, 6EH7,
12,000 I think?  From memory, but right ballpark.

HOWEVER -- and this is the point that matters, through 40M -- at the
same gain level, the 6EH7 will have far less crossmodulation than
other tubes.  It was designed principally with that in mind.

I've substituted 6EH7's in several sets and if you have a halfway
decent audio section, you easily can hear the improvement. Voice
signals are simply clearer.

> ---more snippage---
> 
>> 85 kc R-23 IFTs can be cascaded for better selectivity and shape
>> factor. If you're only going to cascade one pair, yes, put them in the
>> second mixer plate circuit, since that protects the rest of the IF
>> chain from out-of-passband signals that can cause crossmodulation.
> 
> This statement is more important than it would at first appear. You
> want to have the available reciever selectivity as close to the
> antenna as possible, which means doubling up IF transformers at the
> front of the IF strip.
> 
> Cascading the transformers at the 1st IF strip is at least as
> important, and a sharp preselector is more important still.

All these points are important.  Receivers can (and do) overload in
every stage up to and including the audio section.  Generally
speaking, however, the signal is getting larger as it goes, so it's
important that there be steadily fewer unwanted signals to combine and
drive a stage out of the linear range.

There's a limit to what can be accomplished: Even really good front
end coils don't go much above Q = 200, with good toroids usually
slightly superior to good air wound solenoids and a bit more so
compared to coils on plastic forms. So the half-power bandwidth on 80
meters is going to be no better than about 20 kcs and proportionately
more on higher bands.  40's often the worst because of the high power
SW broadcast stations.

40 was going to be rearranged to reduce the mess; I don't know what
happened to that.

<SNIP>

>> > I had intended to use a pair or even 3 6EH7's in my IF strip, and running
>> > cascaded pairs of these transformers at the first 2 coupling points. That 
>> > would be 6 transformers total. But now I'm thinking that a strip with 3 or 4 
>> > 6BJ6s would be easier to tame. I'll breadboard both and run some tests, I 
>> > suppose.
>> 
>> The 6EH7 is a superior tube because it was designed for IF service at
>> up to 40 Mcs in premium TV's where extremely large signals have to be
>> handled without crossmodulation. It's really the only pentode that I'm
>> aware of developed for wide-range signal handling ability at HF/low
>> VHF. Two is plenty for an HBR IF section. Three 6BJ7's might be a bit
>> simpler but short leads, careful orientation of the sockets, and good
>> bypassing isn't rocket science, especially at 85 kcs. You DO need
>> larger bypass caps than you might expect -- .05 mfd or more -- at
>> these low frequencies.
> 
> Yeah, I keep going back and forth on the IF strip. Each 6EH7 will get me 
> about 40 db of gain, and I only need about 120 total for the entire receiver. 
> The thought behind using 3 6EH7s was to have plenty for the AGC to work 
> with.

You'll have plenty with just the two, any reasonable mixer, and an RF
stage.
> 
> Now then, we had been discussing how to couple cascaded IF transformers 
> together and someone mentioned that a tube works really well. DUH! But I 
> sure don't want 4 6EH7s! They'd be just dying to go into oscillation. Other 
> candidates in order of highest gain to lowest: 6BZ6, 6BA6, 6BJ6. If you're 
> going to use 4 stages, 6BJ6 will be the most well behaved, still provide more 
> than enough gain, and they use half the heater current than the others. 4 
> 6BJ6s need the same heater current as 2 6EH7s.

At the same gain level, no tube is much more likely to oscillate than
any other and in fact, the tubes that are designed for very high gain
and VHF use have features to make them easier to tame: internal
shielding, short internal leads, and multiple cathode leads for
example.

However it is true that the same amount of gain spread across more
stages is less prone to some kinds of instability -- namely, feedback
from plate to grid of a given stage and the strip input and output are
likely to be farther apart. In exchange, you get more trouble with
multi-stage kinds of instability.

In my version of the HBR series -- same basic plan but used command
receiver RF coils plugged in through the front panel beneath the dial
somewhat like the HROs -- I used a 6EH7 RF stage, a 6DJ8 push-push
mixer, a 6EH7 1st IF at 1665 kcs, a 1750 kcs 2nd oscillator, and a
single 6EH7 2nd IF at 85 kcs with two pairs of transformers with the
coupling set to minimum. There is PLENTY of gain. Doing it again, I
would put both IF tubes at 85 kcs because with so much gain preceding
the second mixer, there is some crossmodulation there, on very crowded
bands with lots of strong signals.

I used a 12AT7 as a 2nd mixer; a 12AU7 might well solve the problem.

This receiver was something of a ship-in-a-bottle exercise. The unique
plug in coil scheme IS very cute and convenient for band changing, but
there's so little space in the cans that temperature compensation
requires itty-bitty parts. I proved it could be done, but wouldn't
recommend it for anyone else.

I did have trouble getting enough RF gain with the low-Q command set
RF coils and some of them were replaced with toroids in the same cans.
But that was self-inflicted: following the W6TC scheme it would be
fine.

Walt Hutchens
KJ4KV