[Patrick Turner]

====================
Alex replies:
I see your approach to radio design. You are prepared to improve a boring
typical aussie radio by adding a cathode follower, another audio stage for
tone control, replace PT, drill/punch a couple of holes for the extra 2
tubes, etc.

What I can not understand is the craziness. Why go through all this bother
if a better improvement can be done with transistors and op-amps, and the
stuff can be hidden on a small board under the chassis.

I've got a lotta tubes and bits and peices. Most SS AM radios are
quite poor performers. I enjoy working with tubes.

it seems fraudelent to get someone to pay to have their radio restored
by fitting a little board full of chippy crappy thinges.

One has to draw a line some place.

Unfortunately there
is no area where tubes outperform solid state, except for (literaly) warm
feeling.

Try telling that to ppl who prefer tube gear. It just is not true for
them.

I have a Marantz AM-FM chip based tuner sitting on top of my AM radio.
I can switch to FM if I want to, and the FM is not better sounding
than my tube AM set.
When I switch the tuner to its AM section its much inferior to the
tubes, so I use the tubes for AM and audio, and the chip tuner only
for the FM signal.
Something like a Leak Troughline FM tuner is excellent, also Quad FM
tuners. But the tubed stereo decoders often increased noise and
distortion so they were not always good. The add-on MPX made for Leak
and Quad with Ge transistors were awful. OK, so a chip based MPX Does
Work OK. I once instaled a kit made in the UK into an old Kenwood AM-
FM receiver with mono FM - worked very well, and there wasn't a tubed
MPX in there already.

Sometimes I do redesign as well. For example, the best AM detector is the
emitter detector. It virtually combines your cathode follower (emitter
follower in this case) with a diode detector (emitter junction). It has
rather high input impedance: beta times higher than of a bare diode
detector. And beta is quite high for the modern transistors. One problem
though. Reverse breakdown voltage of an emitter junction is about 8V, so the
carrier shall not exceed 1Vm (with a margin). Thus the radio needs to have a
perfect AGC.

Recently I used a couple of small bjts in darlington pair EF mode to
buffer the signal after Ge diode and RC - worked very well to avoid
cut off distortion because of AC loading. But I prefer 1/2 a 12AU7.

An AGC in a boring radio is bad. On a week station AM detector gives 0.5V,
while on a strong local station can reach 5...10V. What I do, I put an
integrator in the AGC circuit so that no matter how strong is a station, the
carrier level is about 1V, while the AGC can be anywhere from 0V to --30V.

Yes, but today I finally got a re-built 1935 radio working fairly well
with 6AN7 triode-hexode mixer and 6N8 IF remote cut off IF amp. To
avoid the terrible hum when tuned to AM stations ( due to compact
fluorescent lamps) I replace the existing RF input coil designed for
antenna = 3 metres of wire. The lamps tend to modulate the
electrostatic content of the electromagnetivc wave at the rectifier F,
100Hz plus horrid harmonics. The replacement is a ferrite rod antenna
which reacts to the magnetic portion of the wave which seems to have
far less hum. But the voltage level is smaller at the input. But with
only 2 tubes for the AM tuner, 4 out of 7 local AM stations produce
between -5 and -8 Vdc of AGC voltage, and audio level does not change
much for the stations even though I have no DC amp to assist the AGC
function.

On my own AM radio I have IF amp = 6BX6 which is sharp cut off and I
have local current FB to linearise the IF amp. AGC is applied to an RF
input stage and mixer only, sure the AF levels vary more but it sounds
better.

I've done a few sets which have had all octal tubes, 6U7 RF, 6A8
mixer, 6U7 IF amp, and because you have THREE stages and Gm is only
moderate, they work very well indeed, and give better SW than the two
stages with just mixer and IF amp. I've never felt a need for
amplified AVC voltage.

The radio becomes very convenient. Set volume and cruse the dial. All the
stations will have the same volume. The radio does not blast when running on
a strong local station.

Nor does mine, and I am not alergic to changing volume for the mood,
to turn up music and intelligent discussions, and to trun down the
crap, adds and lying politicians. Moderate AGC is fine for me.

Now I ask you: how would you make an integrator on a tube? Possible, but
cumbersome. Why lika communist, create difficulties and then overcome them
with a loud chest-pounding and shouting slogans?

What's an integrator? Last time I looked, it was something than
changed a square wave into a saw tooth wave, ie, RC circuit. Unlike
the Chest Pounding Communists, I am no slave to any unruly mob called
'Socialists For Electronic Correctness'. I'll do things my way, or a
better way if I am convinced its yet another intelligent use of a tube
or two, or three.

Well, normally I do not do deep redesign, but just improve whatever possible
keeping the same topology.

I find it impossible to retain existing circuits when I know that I
know better. It is a private thing with me. There are gory details.
But usually the music survives better than otherwise.

That is where the idea of low cut off in the
feedback came from. Once you begin to apply NFB to a lousy boring amp (to
reduce the speaker boom and make crispier sond), you want not to extend the
low frequencies below what the lousy OPT can handle naturally.

But what if the amp is NOT "lousy boring amp" and what if there is NO
speaker boom,
and no need of the extra NFB at VLF?

Feel free to add the cap if you want but I have yet to explore the
idea and make careful measurements to verify to myself the idea is
worth persuit. So far nobody seems to have devoted a website page with
all the details of performance which is too difficult to explain here
in words only.

And you can
not use RC high-pass filters on the input because the input impedance must
be 10M or more (not to load the detector). And you can not use a Hi-Z RC
filter of say 470pF and 10M, as hum and noise caused by flickering grid
current would be enormous.

Sure. I get around all the problems you tell me about by abolishing
the use of grid leak biasing, and using something different. Today I
spent yet another 8 hours on a 1935 radio in which all old tubes were
removed and replaced with 1950s tubes. There had been a mix of 3
oddball European tubes wth Queer Sockets plus a couple of octals made
in Oz; someone had already modified the set, and done it quite badly,
and what was there in 1935 isn't clear, but basically just mixer, IF
amp with diodes, AF input, AF output and type 80 rectifier, and a
Magic Eye. 6 tubes. Now it has 7, with Si diodes to replace 80. I have
yet to re-wire the magic eye. The power lost in the 80 is now used in
the EL34. I found that instead of triode on this set I could have a CT
on the OPT primary for 50% SE UL mode and 12dB GNFB works fine. B+ is
+375V at the same Ia as used for a 6F6 which had been used. The EL34
in SE UL or SE triode gives far nicer sound than 6F6, 42, 6V6, EL84,
and other crap in pentode/tetrode mode with very little GNFB. The only
down side is that when Rout is high with pentode mode the small amount
of treble produced in old radios is "naturally boosted" by the rising
signal at the speaker at higher F because the speaker impedance rises
with F. Some makers were clever to marry the output tubes to just the
right speaker to give a boosted treble. If GNFB is used the treble
dissappears. So then one must use a tweeter. But most old sets without
GNFB become tiring to me and where there should be clear HF there are
competing high levels of "artificial HF", ie, the high levels of THD
and IMD - Its OK if all you need to know is the latest news bulletin,
but not so hot for music.

The set I'm working on will have 2 RCA sockets to combine L&R inputs
from an FM tuner or CD player, ipod, or other i-crap.

Patrick Turner.