[HBR] Gain in IF amplifiers

[Walt Hutchens]

Jim said:
> I suspect that the real problem with the hopped-up S-38 is that the layout 
> and bypassing were simply not adequate for the job of all that extra gain.

I'm inclined to agree with Jim's analysis.   

Building stable multi-stage IF amplifiers is one of those things 
that's usually easy if you do it right and darn near impossible 
otherwise.   With just one IF stage and that probably a 12SK7 (Gm 
= 4000?) an S-38 would have almost no attention to the details of 
how IF stages should be laid out or wired and the thought of trying 
to go back and put that attention into an existing chassis makes 
me faintly ill.   

(There's an interesting project though -- design a hollow state all 
band ham receiver using five tubes.  Keep the design low cost but 
beat the S-38 performance as much as possible.  I think there are 
radar IF tubes with Gm in the 22,000-30,000 range; they ought to 
be okay in single stage receiver IFs.   Add a beam deflection mixer 
with the oscillator in the same tube ... Hummm ... seems like you 
could even do double conversion in five tubes.   Osc/1st mixer, 
Osc/2nd mixer, IF,BFO/prod. detector,1st/2nd Audio -- But no 
provision for AM reception ... that would need work.

(Radar IF stages are stagger tuned in pairs or triplets so the per-
stage gain and associated stability issues aren't nearly what they 
might seem with extremely high gain tubes.   I had one chassis 
with six stages that was spec'd to be flat from 20 Mcs to 40 Mcs.)

(Well the 'modern S-38' is for another time.)

Let's see how many of the rules for IF layout and construction I can 
write down, based on mistakes I've made within (say) the last two 
years.

Everything from the plate of the mixer to the input of the second 
detector is part of the IF strip and must be laid out and wired 
accordingly.

IF strips are best laid out in a straight line.   Even a right angle 
bend makes things harder.  

Keep the area around the input to the 1st IF absolutely clean -- 
nothing unnecessary is allowed there.   A shield over this area can 
be helpful in cases of particularly high gain or when using a high-
impedance low loss filter at this point.  

A shield between the input and output ends of a crystal filter is a 
good idea.   The better the filter, the better the shielding you need 
to avoid degrading both the shape and the ultimate rejection.

If you use a dual-section tube (triode-pentode ...) in an IF stage 
there are strict limits on what can be done with the non-IF section.  
Any RF there, however slight, WILL be coupled to the other 
section, you can count on it.   I have successfully used triode 
'other sections' as rectifiers,  AGC detectors (sharing with the 
stage that drives the detector), and audio amplifiers.   Other things 
failed.

Use only tubes with basing that provides good isolation between 
control grid and plate -- basically that means the tube has to be 
intended for something like this sort of work.   Most pentodes are 
but if you go off the beaten path, be careful.

Orient sockets for short grid and plate leads.

Modern miniature tubes -- anything with Gm >5000 or so -- need a 
VHF parasitic suppressor.  I generally put a 100 ohm 1/4W resistor 
in series with the grid lead with a lead not exceeding 1/8".   If you 
omit this, these tubes can oscillate well into the low VHF 
(frequency determined by lead lengths and stray capacitances) and 
the first time you troubleshoot this problem you'll take the pledge 
never to go there again.   

One ground per tube, directly to the chassis.   I usually put a lug 
under the socket mounting screw nearest the plate lead.   All 
grounds and bypasses after the grid of the tube and up to/including 
the grid of the following stage go to that one lug.   Examples: 
filament, cathode, screen grid, plate supply, and following AGC line 
bypasses.

You can't beat the modern small disc ceramics for IF stage bypass 
work.   .01 mf is fine for most HF IF's but I'd look for small .05's for 
a 455 kcs IF, at least for the cathode and plate decoupling.  

The IF tuning caps should be internal to the transformer.   

Everything on the socket that can be grounded, must be.   Center 
pin for miniature tubes, internal shield if provided, a filament pin if 
possible, any spare pin on the socket, and so on.  If there is more 
than one cathode lead, bypass them both.

IFT's are located between the tubes.   They must be well shielded 
and shields well grounded; be careful with used parts that crud 
doesn't interfere with shield grounding.  If you have a crystal or 
mechanical filter it's handled the same way as an IFT.   

Space tubes and IFTs enough for ease in installing/removing 
shields but not more than 1/8" more than that.   

Don't run power or any other wiring closer than 1" to the edge of the 
line of sockets; 2" is better.   Bring in necessary supply leads 
directly at right angles to the line of sockets.   Ditto decoupling, 
and cathode resistors -- try not to run small parts lengthwise along 
the IF strip.   I put lug strips 1-2" off to the side from each socket, 
connect decoupling resistors from there (keep leads as short as 
practical at the socket end) and run the various busses along 
outside of the lug strips.    

Watch out for coupling along the AGC line -- cheap receivers with 
low stage gains can use little or no decoupling but with a pair of 
6BZ6's or 6EH7's -- no.   

All leads must be short and direct.   

IFs with unusually high gain may need a shield mounted across 
each socket.   

All of these rules can be violated (at least one or two at a time) and 
in the right circumstances you'll get away with it.   But when 
building new, it's a whole lot easier to follow them than it is to do 
otherwise and then try to go back and conform when the IF is 
unstable or the filter bandpass shape varies with the AGC voltage.

Walt Hutchens
Trashing for a pound what others fix for a shilling, for over 60 years.
KJ4KV