[Big Bad Bob]

Patrick Turner wrote:
Let us consider detection of AF from an IF amp producing a 455kHz IF
signal.

Low signal levels fed into any sort of old fashioned diode detector
with 1 diode plus C plus R will always produce serious distortion
levels at high modulation % regardless of the mudulation F because of
the high ripple which may be present if the time constant for the CR
filter after the diode is suitable to be able to get an undistorted
high output AF voltage, say 10Vrms at say 10kHz.

very high impedence in the detector circuit would help offset this. That
and an LC filter. But you're right, even a diode has nonlinearity.

"Bottoming out" that you see with high mod % could be avoided if the
effective mod % could be reduced to say 30%, when the IF signal is at
100%, by means of adding a cinstant level of 455kHz in correct phase
relationship to the IF signal.

This complicates the circuit, though, doesn't it? I'd favor a sharp cutoff
pentode biased near its cutoff point, and with a relatively small gain (Rp /
Rk 10) as the detector (there could be a capacitor across Rp to filter the
IF). You should be able to handle close to 100% modulation this way, with
minimal THD.

One solution COULD BE to employ a synchronized or locked oscillator
which produces a constant level of 455kHz signal which is 180degrees
out of phase to the IF wave being detected

PLL - ok. Even a crystal oscillator can become a PLL if the phase
discriminator output is large enough.

But if I had to make a PLL I'd use a quadrature detector. And a crystal
filter on the 1st IF stage. And it could be done with tubes, but you'd need
quite a few of them. Some of the "high end" comm equipment that the U.S.
Navy had "back in the day" actually did things like this. And they were
pretty much "all tubes" with crystal ovens to keep frequencies stable.

[Patrick Turner]

On Nov 25, 9:25*am, Big Bad Bob BigBadBob-at-mrp3-
wrote:
Patrick Turner wrote:
Let us consider detection of AF from an IF amp producing a 455kHz IF
signal.

Low signal levels fed into any sort of old fashioned diode detector
with 1 diode plus C plus R will always produce serious distortion
levels at high modulation % regardless of the mudulation F because of
the high ripple which may be present if the time constant for the CR
filter after the diode is suitable to be able to get an undistorted
high output AF voltage, say 10Vrms at say 10kHz.

very high impedence in the detector circuit would help offset this. *That
and an LC filter. *But you're right, even a diode has nonlinearity.

But high Z circuits suffer from stray C effects. Usually most tube
radios are desiged so local AM stations generate about -4Vdc AVC
voltage which means that the carrier level is maybe 5Vpeak, so the
maximum recoverable AF is high enough to be a lot greater than the
diode threshold. Detectors are usually worse and low level signals and
best when high signals are present, but not too high, less slew
distortion occur.


"Bottoming out" that you see with high mod % could be avoided if the
effective mod % could be reduced to say 30%, when the *IF signal is at
100%, by means of adding a cinstant level of 455kHz in correct phase
relationship to the IF signal.

This complicates the circuit, though, doesn't it? *

Indeed it would. Its applying the principle of adding an elavated
carrier to make a low level signal which may have modulation between
0% and 100% look like a signal with much less % modulation and which
may then be detected more linearly, and with natural selectivity and
lower noise benefits.

I'd favor a sharp cutoff pentode biased near its cutoff point, and with a relatively small gain (Rp utoff
pentode biased near its cutoff point, and with a relatively small gain (Rp /
Rk 10) as the detector (there could be a capacitor across Rp to filter the
IF). *You should be able to handle close to 100% modulation this way, with
minimal THD.

I have not tried this.

One solution COULD BE to employ a synchronized or locked oscillator
which produces a constant level of 455kHz signal which is 180degrees
out of phase to the IF wave being detected

PLL - ok. *Even a crystal oscillator can become a PLL if the phase
discriminator output is large enough.

But if I had to make a PLL I'd use a quadrature detector. *And a crystal
filter on the 1st IF stage. *And it could be done with tubes, but you'd need
quite a few of them. *Some of the "high end" comm equipment that the U.S.
Navy had "back in the day" actually did things like this. *And they were
pretty much "all tubes" with crystal ovens to keep frequencies stable.

Maybe super selective, but low AF bandwidth.

What I like in an old radio is AF BW at least 7kHz, and yet have
excellent skirt selectivity to keep out other nearby stations on the
BCB.

Usually it can be done minimally using very carefully set up RF input
coils with good tracking with the VFO, and then a pair of IFT with
first one having a tertiary winding to make the pass band wider, with
slight twin peaked nose response which then combines with the broadly
peaked of the second IF to give an overall response that is +/-7kHz
each side of 455kHz. In the 1960s many imported Japanese AM radios
had such a feature in receivers which also had stereo FM and two good
PP audio amps. But ordinary local made AM radios didn't, and they had
only 3kHz of audio BW. Quad also had such a wide band feature for AM.
Clever dicks can wind on a few more turns for a tertiary in an IFT to
make its BW wider.
My own set I made 12 years back has one IF coil on a slide so that the
critical coupling may be increased or decreased. Halicrafters had a
similar feature on their sets, with 3 IFTs all with variable coupling.
I got my idea from Hali. But the idea of the tertiary is easier to do
and is explained in RDH4.

Patrick Turner

[Big Bad Bob]

Patrick Turner wrote:
What I like in an old radio is AF BW at least 7kHz, and yet have
excellent skirt selectivity to keep out other nearby stations on the
BCB.

I was once given an old 'hi fi' (heathkit) set which was all compoments,
dated around 1955. Unfortunately it's all gone now, for various reasons. I
still remember what they did in the AM radio, though - they tweeked the IF a
bit so it had better bandwidth (~10khz) and put some peaking coils and LP
filters in the AF stage to boost up 10khz and drop out 12Khz (or similar).
I can't remember if they used the sharp cutoff pentode detector or tube
diodes, but there weren't any semiconductors in it. The FM receiver DID use
tube diodes with a classic ratio detector setup.

Usually it can be done minimally using very carefully set up RF input
coils with good tracking with the VFO, and then a pair of IFT with
first one having a tertiary winding to make the pass band wider, with
slight twin peaked nose response which then combines with the broadly
peaked of the second IF to give an overall response that is +/-7kHz
each side of 455kHz. In the 1960s many imported Japanese AM radios

the IF bandwidth increase on this radio was done by dual-tuning the IF
transformers, with the primary tuned on one side of 455khz, and the
secondary on the other (this practice is also common in older TV sets). The
tuning slugs were on opposite ends of the transformer, so coupling should
not have been affected (much) by tuning. That's one reason why the IF
transformers are so large.

Nowadays a decent crystal filter will give you a nice bandwidth patern by
doing the same thing, with 3 crystals at different frequencies around 455khz
forming a 'pi filter' (I think that's the right name). That and no tuning
required.

[Patrick Turner]

On Nov 25, 7:24*pm, Big Bad Bob BigBadBob-at-mrp3-
wrote:
Patrick Turner wrote:
What I like in an old radio is AF BW at least 7kHz, and yet have
excellent skirt selectivity to keep out other nearby stations on the
BCB.

I was once given an old 'hi fi' (heathkit) set which was all compoments,
dated around 1955. *Unfortunately it's all gone now, for various reasons. *I
still remember what they did in the AM radio, though - they tweeked the IF a
bit so it had better bandwidth (~10khz) and put some peaking coils and LP
filters in the AF stage to boost up 10khz and drop out 12Khz (or similar).

The tweaked IF could have been an overcoupled first IFT with coils
about 10mm closer than usual for the normal wide single peaked
response used normally for both IFT. And in some radios, especially
portables the IFTs went after selectivity with a narrow bass band, ie
hi Q. AF BW was usually poor in portables but gee you looked cool on
the beach in 1947....


* * I can't remember if they used the sharp cutoff pentode detector or tube
diodes, but there weren't any semiconductors in it. *The FM receiver DID use
tube diodes with a classic ratio detector setup.

The commonest approach is to have a mixer, say 6BE6, followed by
variable µ pentode for IF amp which has 2 diode anodes which allow one
for a negative going AF voltage and the other is used to generate AVC
using a 33pF cap off the IF tube anode. This is the best way to get
AVC because the anode tank has slightly wider IF BW than the secondary
LC tank. So AVC tunes more broadly, and can be a bigger -Vdc signal
and it tends to lessen tuning hash noise before and after a station is
tuned in right. The two diodes sometimes appear in a triode which is
the first AF tube. But not in reflexed sets which use a dual diode-
pentode for IF amp then also for the "audio run around again" after
its been detected. Such a set saves a tube, and is suitable for AF
shunt NFB from OPT sec and including the volume control. I have a
quite fine sounding set which manages to get the IF tube to wear two
hats while being cajoled into singing rather than screaching by the Mr
NFB and his genteel whip.

But I still don't like reflex. They can become unstable.

My own best AM radio has a vary-µ pair of triodes with AVC applied as
a cascode RF amp, 6AN7 mixer, 6BX6 as the IF amp which is sharp cut
off and fairly linear, then 12AU7 triodes used for CF to drive Ge
diode detector plus R&C with almost CCS drain from the C to prevent
slew, and a following buffer stage and treble control with +/- 6dB of
boost abd cut. Then the AF amp is EL34 in triode driven by 12AX7
parallel. IF BW is 9kHz max, and the local AM stations are allowed to
transmit 9kHz, so get everything, and its little different to
listening to an RF station.

I did have dual cascaded RF input coils slightly stagger tuned but I
changed to a ferrite rod because electrostatic noise from compact
flouro lamps became intolerable. So I mainly get the magnetic portion
of the electromagnetic RF wave and it has less noise.

* Usually it can be done minimally using very carefully set up RF input
* coils with good tracking with the VFO, and then a pair of IFT with
* first one having a tertiary winding to make the pass band wider, with
* slight twin peaked nose response which then combines with the broadly
* peaked of the second IF to give an overall response that is +/-7kHz
* each side of 455kHz. *In the 1960s many imported Japanese AM radios

the IF bandwidth increase on this radio was done by dual-tuning the IF
transformers, with the primary tuned on one side of 455khz, and the
secondary on the other (this practice is also common in older TV sets). *The
tuning slugs were on opposite ends of the transformer, so coupling should
not have been affected (much) by tuning. *That's one reason why the IF
transformers are so large.

In practice wonder about this form of stagger tuning. Its hard to get
right have found, and ppl sometimes twiddle cores for max AF output,
and when all are centre tuned the AF response becomes accumulatively
narrower.

Not nearly as bad as the single tuned coils used in early AM radios
with bjts. Typically there was only a total of 3 tuned circuits, the
ferrite rod coil and two "IFTs", and nowhere near enough skirt
selectivity was present where a stron local was say only 50kHz away.
(( Skirt selectivity is a technical radio term and is not to be
confused with "hunting for crumpet" in the bars and clubs on Saturday
night , :-) ))

Nowadays a decent crystal filter will give you a nice bandwidth patern by
doing the same thing, with 3 crystals at different frequencies around 455khz
forming a 'pi filter' (I think that's the right name). *That and no tuning
required.

Yes, but I have not tried all that. Usually the right kind of X-tal
filters are not around to be bought, and then if they are, they need a
low impedance drive, and so the IFT looks right with a TOOB.

Patrick Turner.

[Patrick Turner]

On Nov 30, 1:44*pm, "Alex" wrote:
"Patrick Turner" wrote in message

news:32a07bea-4121-495b-

My own best AM radio has a vary-µ pair of triodes with AVC applied as
a cascode RF amp...

Alex:
What are these vari-mu triodes? 6ES8? or just 6BA6 triode connected?

I just had a look at the schematic I drew about 10 years ago and the
cascode triode hs not been labelled.

Its a twin triode though, with AVC applied. I had two LC input
circuits slightly stagger tuned with 39k between tops of LC for
coupling, and the second coil went to grid and 005uF where AVC is
applied. The anode load on the top tube is 22k resistance, CR coupled
to grid of 6AN7 mixer.

I've since installed a ferite rod antenna because a long wire picked
up far too much noise from compact fluorescent lamps despite heavy
legislation against any device capable of RFI.

But I don;t remember the tube number.

There are not many vari-µ triodes, and a pair of trioded 6BA6 could be
used. Perhaps I was overzealous with a cascode for low noise on the
BCB, and a normal vari-µ pentode in pentode mode such as a single 6BA6
would be just fine because the signal level is so low the distortion
will be low.

Patrick Turner.

[Alex]

"Patrick Turner" wrote in message
news:32a07bea-4121-495b-

My own best AM radio has a vary-µ pair of triodes with AVC applied as
a cascode RF amp...

Alex:
What are these vari-mu triodes? 6ES8? or just 6BA6 triode connected?