AM detectors, was re: OPT equalising circuit

Alex wrote:
> 
> "Patrick Turner" <info@[EMAIL PROTECTED]
> wrote in message
> news:481C8DCE.3448A31C@[EMAIL PROTECTED]
> > > > I repair, rewire, or seriously modify at least 5 AM radios per
year.
> > > > I've done 3 this year and there are 3 waiting to be done, and its
only
> > > > May.
> > > >
> > > > I built a complete radio in 1999, 10kHz of AF bandwidth, variable
> > > > selectivity, linear IF amp, linear detector, linear AF amp with
> > > > EL34 in triode with 12AX7 and 12dB FB. Rola 12" Deluxe speaker
from
> 1953
> > > > with a dome tweeter from 1972, because the 12" speaker is fairly
flat
> to
> > > > 5kHz.
> > > >
> > > > I have never heard any other AM sets that come close except
> > > > the one in the AM-FM tuner at my website, which I designed, and of
> > > > course the
> > > > Quad AM tuner all tubed, that isn't bad at all.
> > >
> > > Could you please give a link to your website? Is there a schematic
of
> that
> > > radio, mentioned above, which you designed? Is it TRF?
> > > Regards,
> > > Alex
> > >
> > > > Patrick Turner.
> >
> > Try
> >
> > http://www.turneraudio.com.au/am-fm-radio-tuner-multiplex-decoder.html
> >
> > Patrick Turner.
> Interesting.
> My comments:
> 1. I am not that fanatical to build a stereo-decoder around LC tanks
which
> require precise tuning.

But becoming an expert in servicing such things requires that you design
and build
such a thing as you would service. Its called doing an apprenticeship. 

> 2. It was funny to see a reactive tube for AFC instead of a tuning diode
(or
> any diode like 1N4148).

no problem though; the tube works well...

> 3. As far as the AM detector is concerned, I think it is not optimum.
> Voltage on the detector appears to be about +35V. Forward bias current
is
> about 90uA. Differential impedance of the diode is about 250ohm (no
signal),
> 220pF capacitor impedance is about 1.5K at 455KHz. Thus at no signal the
CF
> is loaded by about 1.5K. The CF, running at 1mA DC current would have
about
> 0.5mA/V transconductance, therefore output impedance is about 2K. Thus
the
> gain of the CF is only 0.4 or so.

I was able to get a very much larger linear output 
from the CF detector than is required.

It works, try it. The CF idle dc current of 1mA is OK.
But the gm is much higher than 0.5mA/V.
As the tube turns on harder gm much increases.
It wouldn't work well to get a negative going voltage.
But OK for a positive going voltage.
THD is well under 1%, better than 5% as in most detectors.
 


> 
> For the detector to begin detecting, obviously the AC current through
the
> diode shall exceed the DC component (90uA). It will require about 150mV
on
> the input to the detector or conversely 400mV on the grid of the CF.
> Sensitivity of such a detector is rather poor (as you know, sensitivity
of a
> pure p-n diode is about 25mV.)

Doesn't matter because the signal to be detected from the IFT is so
large
that one can get about -12Vdc of AFC voltage, more than needed.

> 
> It is easy to improve the result 10 times -- just reduce the bias.
Connect
> R17 not to the ground, but to point "R" (+31V). After that the diode
will be
> biased only by the self bias of the tube, which is about 4V. Sensitivity
of
> the detector will increase 10-fold.

The sensitivity you mention isn't needed or wanted.

The R16 and R17 network are 330k and 100k = 430k,
as as the cathode of CF is at about +33Vdc, there is a dc current
through the diode = 33/430 mA
or about 0.09mA, and this stays fairly constant with changes to AF
voltage in the 220pF.
I found having a virtual constant current discharge of this C was much
better than having 
a large R taken to a voltage not much different to Ek.

I suggest you build this detector and give it a decent test using a
455kHz modulated test
signal. The main advanatge is that at high percentages of modulation
used these days the distortion
remains low until very close to 100%. 

In many radios the detector suffers cut off distortion as % mod rises.

With very small signals, the detector suffers no diode threshold effects
because the 
diode is 'on' at all times.

Patrick Turner. 
 
> 
> Regards,
> Alex