In article , Robert Casey
wrote:

Simulated and actually built the below AM detector. Uses two cathode
followers
and two diodes. One a 12AU7 and the other a 6AL5 with low heater voltage
(makes for lower contact potential and better linearity). The
simulation says
distortion is about 50 dB down for input carrier levels of 20Vp-p to a
low of 20mVp-p.
This is by far the best I've seen. And the hardware I built (modified a
Heathkit AM tuner)
sounds great. The circuit is similar to PT's except I used 100pF caps
to couple the
CF'ed carrier to the detector diodes (one for audio, other for AVC).

If the schematic you posted is correct, then you seem to have eliminated
the bias circuit from your detector, which makes sense, because as the
RDH4 bit that John Stewart pointed out says, the point of the bias is to
compensate for a less then optimal AC/DC load ratio, and with the
following cathode follwer you seem to have an ideal AC/DC load ratio.

I don't yet understand the claim that Patrick has made that the bias
insures that the diode is "on all the time", or that you have made that
the bias is to "get it up above the Knee", I would welcome further
enlightenment on these aspects of the design.

As it is, it sounds like you may have equaled the distortion performance
of the Selsted & Smith detector. It would be interesting to learn how
your detector performs when the first cathode follower is removed, and the
diode is driven directly from the IFT. That would yield a diode detector
buffered only on the post detection side by a cathode follower, which is
my first choice for a meritorious detector.

Speaking of the IFT, I would tend to dismiss reports of how your circuit
sounds in the Heathkit tuner, unless you have provided a resistive load
for the secondary of the IFT equal to the load presented by the original
voltage doubler detector in the Heathkit. I think the voltage doubler
presents a load to the IFT equal to about 1/4 of the original detector
load resistance. Also your circuit is going to provide less AGC voltage
than the original circuit, and for an apples to apples comparison
something should be done to address that difference.


Regards,

John Byrns


Surf my web pages at, http://users.rcn.com/jbyrns/

John Byrns wrote:

In article , Robert Casey
wrote:



Simulated and actually built the below AM detector. Uses two cathode
followers
and two diodes. One a 12AU7 and the other a 6AL5 with low heater voltage
(makes for lower contact potential and better linearity). The
simulation says
distortion is about 50 dB down for input carrier levels of 20Vp-p to a
low of 20mVp-p.
This is by far the best I've seen. And the hardware I built (modified a
Heathkit AM tuner)
sounds great. The circuit is similar to PT's except I used 100pF caps
to couple the
CF'ed carrier to the detector diodes (one for audio, other for AVC).



If the schematic you posted is correct, then you seem to have eliminated
the bias circuit from your detector, which makes sense, because as the
RDH4 bit that John Stewart pointed out says, the point of the bias is to
compensate for a less then optimal AC/DC load ratio, and with the
following cathode follwer you seem to have an ideal AC/DC load ratio.


That's correct, I did remove the bias. With the cathode follower
directly coupled to the
detector circuit output, there are no caps other than the picofarad
sized ones. Which
discharge faster than the audio waveform changes. But these caps need
to be chosen with
some care to maintain a decent AC/DC ratio or else the downward stroke
of the audio
will get a "slope clipping".


I don't yet understand the claim that Patrick has made that the bias
insures that the diode is "on all the time",


I don't either. If the diode is on all the time, it would just pass the
carrier waveform and
not detect.

or that you have made that
the bias is to "get it up above the Knee", I would welcome further
enlightenment on these aspects of the design.


Real diodes have a nonlinear curve down near zero volts. A 6AL5
operated with low
heater voltage (about 4V) is the best, but still has some nonlinearity.
Full heater voltage
makes it a little worse, but allows higher peak currents. Geranimium
diodes have a
well known knee at about 300mV and Silicon at 700mV. But the vacuum
tube diode
has the least kinked curve and maintains conductivity to zero. SS
diodes pretty much
go dead around 100mV to 400mV (Ge or Si). With the FCC rule saying a
minimum of
5% carrier at the valleys, the audio modulation spans a reasonally
unkinked range at
any signal strength. Oh, about below 1V signal strength the distortion
goes up a bit (40dB
vs 50dB down). But it doesn't go to bananas like it does in an average
AA5.

Seems that the "knee" I was avoiding before was really a lousy AC/DC
ratio.


As it is, it sounds like you may have equaled the distortion performance
of the Selsted & Smith detector. It would be interesting to learn how
your detector performs when the first cathode follower is removed, and the
diode is driven directly from the IFT. That would yield a diode detector
buffered only on the post detection side by a cathode follower, which is
my first choice for a meritorious detector.


I'll have to try it. One thing the CF does largely prevent is a loss of
signal voltage vs. when the
detector directly loads the IFT. But I did just that when I slapped on
a 100K resistor on the
IFT secondary. I'll have to give it more thought....


Speaking of the IFT, I would tend to dismiss reports of how your circuit
sounds in the Heathkit tuner, unless you have provided a resistive load
for the secondary of the IFT equal to the load presented by the original
voltage doubler detector in the Heathkit.


I approximately did, (forgot to mention it). I used 100K as a load
across the secondary.

Robert Casey wrote:

John Byrns wrote:

In article , Robert Casey
wrote:



Simulated and actually built the below AM detector. Uses two cathode
followers
and two diodes. One a 12AU7 and the other a 6AL5 with low heater voltage
(makes for lower contact potential and better linearity). The
simulation says
distortion is about 50 dB down for input carrier levels of 20Vp-p to a
low of 20mVp-p.
This is by far the best I've seen. And the hardware I built (modified a
Heathkit AM tuner)
sounds great. The circuit is similar to PT's except I used 100pF caps
to couple the
CF'ed carrier to the detector diodes (one for audio, other for AVC).



If the schematic you posted is correct, then you seem to have eliminated
the bias circuit from your detector, which makes sense, because as the
RDH4 bit that John Stewart pointed out says, the point of the bias is to
compensate for a less then optimal AC/DC load ratio, and with the
following cathode follwer you seem to have an ideal AC/DC load ratio.


That's correct, I did remove the bias. With the cathode follower
directly coupled to the
detector circuit output, there are no caps other than the picofarad
sized ones. Which
discharge faster than the audio waveform changes. But these caps need
to be chosen with
some care to maintain a decent AC/DC ratio or else the downward stroke
of the audio
will get a "slope clipping".


I don't yet understand the claim that Patrick has made that the bias
insures that the diode is "on all the time",


I don't either. If the diode is on all the time, it would just pass the
carrier waveform and
not detect.

Try building the detector circuit that I have just posted using two CFs
and a handful of other easily obtainable bits.
Then examine all the working voltages and waveforms with a low capacitance
probe
of a CRO.
Then you'll see how these things work.

Compare the results of my detector design with anything else in "normal" sets.

Patrick Turner.



or that you have made that
the bias is to "get it up above the Knee", I would welcome further
enlightenment on these aspects of the design.


Real diodes have a nonlinear curve down near zero volts. A 6AL5
operated with low
heater voltage (about 4V) is the best, but still has some nonlinearity.
Full heater voltage
makes it a little worse, but allows higher peak currents. Geranimium
diodes have a
well known knee at about 300mV and Silicon at 700mV. But the vacuum
tube diode
has the least kinked curve and maintains conductivity to zero. SS
diodes pretty much
go dead around 100mV to 400mV (Ge or Si). With the FCC rule saying a
minimum of
5% carrier at the valleys, the audio modulation spans a reasonally
unkinked range at
any signal strength. Oh, about below 1V signal strength the distortion
goes up a bit (40dB
vs 50dB down). But it doesn't go to bananas like it does in an average
AA5.

Seems that the "knee" I was avoiding before was really a lousy AC/DC
ratio.


As it is, it sounds like you may have equaled the distortion performance
of the Selsted & Smith detector. It would be interesting to learn how
your detector performs when the first cathode follower is removed, and the
diode is driven directly from the IFT. That would yield a diode detector
buffered only on the post detection side by a cathode follower, which is
my first choice for a meritorious detector.


I'll have to try it. One thing the CF does largely prevent is a loss of
signal voltage vs. when the
detector directly loads the IFT. But I did just that when I slapped on
a 100K resistor on the
IFT secondary. I'll have to give it more thought....


Speaking of the IFT, I would tend to dismiss reports of how your circuit
sounds in the Heathkit tuner, unless you have provided a resistive load
for the secondary of the IFT equal to the load presented by the original
voltage doubler detector in the Heathkit.


I approximately did, (forgot to mention it). I used 100K as a load
across the secondary.

In article , Robert Casey
wrote:

Real diodes have a nonlinear curve down near zero volts. A 6AL5
operated with low
heater voltage (about 4V) is the best, but still has some nonlinearity.
Full heater voltage
makes it a little worse, but allows higher peak currents. Geranimium
diodes have a
well known knee at about 300mV and Silicon at 700mV.

There are those that will tell you that Geranium diodes have no knee or
discontinuity, that they follow a continuous mathematical function that
has no discontinuity or "knee". They say that the appearance of a "knee"
is simply due to the way the curve is commonly plotted. IIRC, if you plot
the curve on the right sort of "log" paper, then the curve becomes a
straight line with no "knee" at all.

But the vacuum
tube diode
has the least kinked curve and maintains conductivity to zero. SS
diodes pretty much
go dead around 100mV to 400mV (Ge or Si). With the FCC rule saying a
minimum of
5% carrier at the valleys, the audio modulation spans a reasonally
unkinked range at
any signal strength.

Are you sure about that the FCC has a 5% rule for normal AM monophonic
broadcasting? I am too lazy to look it up right now, but I suspect that
the 5% rule applies to stereo broadcasting only and that normal AM is
allowed to have negative modulation peaks all the way down to 0% carrier.


Regards,

John Byrns


Surf my web pages at, http://users.rcn.com/jbyrns/