wrote in message
ups.com...

Brian Denley wrote:
How can any audio filter make up for severe distortion?

--
Brian Denley
http://home.comcast.net/~b.denley/index.html
-----------------------------------
Please read the pdf at:

http://www.kongsfjord.no/dl/Audio/On...ures%20Of%20Au
dio%20Distortion%20Of%20Received%20AM%20Signals%20 Due%20To%20Fading%20II.pdf

Lots of nifty formulae and even has FFT trasform screen captures to
show his reasoing.

Terry


Am I reading the nifty formulae wrong? It looks to me like he's deriving
the distortion of a diode detector from the modulation index only. My sense
of these things says that a 50% modulated signal at a tenth of a volt is
going to have much more distortion than a 50% modulated signal at 10 volts.

Frank Dresser

Frank Dresser wrote:

Am I reading the nifty formulae wrong? It looks to me like he's deriving
the distortion of a diode detector from the modulation index only. My sense
of these things says that a 50% modulated signal at a tenth of a volt is
going to have much more distortion than a 50% modulated signal at 10 volts.

Frank Dresser

Very few radios drive the detector with anything near 10V.
The R390 and R392 have the highest diode drive voltages I have
seen and I think they are less then about 3V.

Most modern, IE "solid state", receivers I have measured have less
1V. All that I have seen that use discrete diode detectors as oppossed
to ICs, have farily high AF gain stages.

I didn't post this as an attemp to claim that "Synchronous detectors"
are a hoax,
but to offer another viewpoint that is backed up by what appears to be
valid
engineering to me.

ASCII text is not my choice for this arcane topic because of the great
difficulty
in expressing meaningfull equations.

This is merely another tool to be used in trying to receceive fading
signals.
His filters work much better then I expected. I found that by forward
biasing
the detector in my R2000 I got a much cleaner, ie lower distortion,
signal.
This was difficult to manage over very modest temperature changes.
A full wave "improved AM detector" gave even better results.
http://www.amwindow.org/tech/htm/alowdisdet.htm
A synch detector in an outboard detector gave even better results.

But the simple improved AM detector with a 4000Hz LP filter is a pretty
close
match to the synch detector at 1/100 the effort.

The above link goes into the math, this link starts with simpler math
and may
help the none engineers enter the fray.
http://www.st-andrews.ac.uk/~www_pa/Scots_Guide/RadCom/part9/page2.html
Another unusual but good detector can be seen at:
http://www.pan-tex.net/usr/r/receivers/elrpicamdetect.htm
Tom Holden's Synch detector group has a link to a very detailed math
examination of "detection". I lost the link to that paper so you will
have to ask
Tom or join his group.

And please note Mr. Lankford is not merely slapping a 4000Hz LP AF
filter
in the audio chain, he is offset tunning, with good narrow IF fitlers,
to eliminate
one sideband.

Terry

wrote in message
oups.com...

Frank Dresser wrote:

Am I reading the nifty formulae wrong? It looks to me like he's
deriving
the distortion of a diode detector from the modulation index only. My
sense
of these things says that a 50% modulated signal at a tenth of a volt is
going to have much more distortion than a 50% modulated signal at 10
volts.

Frank Dresser

Very few radios drive the detector with anything near 10V.
The R390 and R392 have the highest diode drive voltages I have
seen and I think they are less then about 3V.

The range is extreme, but not outlandish.


Most modern, IE "solid state", receivers I have measured have less
1V. All that I have seen that use discrete diode detectors as oppossed
to ICs, have farily high AF gain stages.

But I'd expect considerably less distortion at 3V rather than 1V.

And I'd also expect that no radio really uses a square law detector to
detect the audio. Real detectors try to linerize a diode's operation by
lightly loading the detector with a reletively high resistance and trying to
minimize operation in the diode's "square law" area. Both voltage and AC/DC
impedance are important considerations in determing diode audio detector
distortion.

I suspect the term "square law detector" is the same sort of term as "first
detector" -- what's now known as a mixer.

I know I've been tripped up by these archaic terms before.

Frank Dresser

In article
,
"Frank Dresser" wrote:

wrote in message
oups.com...

Frank Dresser wrote:

Am I reading the nifty formulae wrong? It looks to me like he's
deriving the distortion of a diode detector from the modulation
index only. My sense of these things says that a 50% modulated
signal at a tenth of a volt is going to have much more distortion
than a 50% modulated signal at 10 volts.

Frank Dresser

Very few radios drive the detector with anything near 10V. The R390
and R392 have the highest diode drive voltages I have seen and I
think they are less then about 3V.

The range is extreme, but not outlandish.


Most modern, IE "solid state", receivers I have measured have less
1V. All that I have seen that use discrete diode detectors as
oppossed to ICs, have farily high AF gain stages.

But I'd expect considerably less distortion at 3V rather than 1V.

And I'd also expect that no radio really uses a square law detector
to detect the audio. Real detectors try to linerize a diode's
operation by lightly loading the detector with a reletively high
resistance and trying to minimize operation in the diode's "square
law" area. Both voltage and AC/DC impedance are important
considerations in determing diode audio detector distortion.

I suspect the term "square law detector" is the same sort of term as
"first detector" -- what's now known as a mixer.

I know I've been tripped up by these archaic terms before.

I'm not a radio circuit designer but detectors circuits are designed for
a certain situation and will not produce the expected output if the
expected input conditions do not exist. All RF carrier and sidebands
(tones) are an alternating wave forms. To recover the AM modulated
information the sideband tones are rectified and averaged, which is the
low frequency audio modulation. The sideband tones are usually much
lower than the carrier but the detector rectifies all of these signals.
For the detector design a minimum signal level is required for it to
rectify the side band tones and the designs have depended on the carrier
to be there so that the detector is switching on and off into the liner
region of the diode. If the carrier is not there then the sideband tone
signal is switching the diode on and off resulting in a lot of
distortion.

The sync detection uses a PLL circuit to lock a local oscillator to the
received carrier and that is summed with the received carrier and side
band tones so that when the received carrier disappears due to selective
fading the locked local oscillator signal is enough to keep the detector
operating in its liner region with just the side band tones present.

The same thing happens using a BFO or when you switch to SSB mode on a
radio but here the local oscillator is not locked to the received
carrier and you have to tune the radio very carefully to get it spot on
the received carrier frequency so the side tones are reproduced at the
original modulation audio frequencies.

Before sync detection circuit designers would use diodes with smaller
non-liner switching regions using germanium for example with lower
forward voltages. These diodes would need less signal power to turn on
and off into the liner region of it operating curves so less energy from
the carrier would be needed to keep the detector in its liner region.
This is a help when the received carrier only fades a little but does
not help if fades a lot or disappears.

Some detector designs would use a DC bias on the diode to put it on the
edge of its liner region to improve its small signal sensitivity. The
optimum bias voltage will depend on the diode characteristics.

--
Telamon
Ventura, California

"Telamon" wrote in message
...

[snip]


Some detector designs would use a DC bias on the diode to put it on the
edge of its liner region to improve its small signal sensitivity. The
optimum bias voltage will depend on the diode characteristics.


There's a linear region in the usual model of a semiconductor diode (a fixed
voltage drop with a series resistance), but that model is only an
approximation. The other model, the square law model, is also just an
approximation, although it's supposed to be close enough over small parts of
the curve.

However, the diode doesn't have to be linear in order to have a fairly
linear diode detector circuit. Imagine we have a diode whose forward
resistance drops in a square law with the voltage. At 0.1V the forward
resistance is 1 meg. At 0.2V the forward resistance is 1K. At .0.3V the
forward resistance is 32 ohms. At 0.4V the resistance is 5.6V, and so on.

Now, let's put this nonlinear diode in series with a linear load resistance
and decide that the circuit is pretty much linear once the diode resistance
drops to 10% of the load resistance. Well, it's obvious that diode detector
circuits which work into higher resistance loads will linearize themselves
at lower voltages than diode detectors which work into lower resistance
loads.

Below a certain voltage, the diode's non linear characteristics will
dominate the detector. Low voltage signals will have much more of their
waveform in this funky reigion than high voltage signals, even at the same
modulation index.

So, as I see it, there's alot more to know about a diode detector's audio
distortion than only the modulation index. There's the actual
characteristics of the diode, the resistance of the load and the signal
voltage the detector is operating at.

There's also the RF filtering, which will tend to "sawtooth" the audio a
bit, much as the rectifier and capacitor do in a power supply. There's also
some resistances/capacitances in the AVC line.

But I could be wrong. If so, let me know!

Frank Dresser

In article ,
"Frank Dresser" wrote:

"Telamon" wrote in
message

.com...

[snip]


Some detector designs would use a DC bias on the diode to put it on
the edge of its liner region to improve its small signal
sensitivity. The optimum bias voltage will depend on the diode
characteristics.


There's a linear region in the usual model of a semiconductor diode
(a fixed voltage drop with a series resistance), but that model is
only an approximation. The other model, the square law model, is
also just an approximation, although it's supposed to be close enough
over small parts of the curve.

However, the diode doesn't have to be linear in order to have a
fairly linear diode detector circuit. Imagine we have a diode whose
forward resistance drops in a square law with the voltage. At 0.1V
the forward resistance is 1 meg. At 0.2V the forward resistance is
1K. At .0.3V the forward resistance is 32 ohms. At 0.4V the
resistance is 5.6V, and so on.

Now, let's put this nonlinear diode in series with a linear load
resistance and decide that the circuit is pretty much linear once the
diode resistance drops to 10% of the load resistance. Well, it's
obvious that diode detector circuits which work into higher
resistance loads will linearize themselves at lower voltages than
diode detectors which work into lower resistance loads.

Below a certain voltage, the diode's non linear characteristics will
dominate the detector. Low voltage signals will have much more of
their waveform in this funky reigion than high voltage signals, even
at the same modulation index.

So, as I see it, there's alot more to know about a diode detector's
audio distortion than only the modulation index. There's the actual
characteristics of the diode, the resistance of the load and the
signal voltage the detector is operating at.

There's also the RF filtering, which will tend to "sawtooth" the
audio a bit, much as the rectifier and capacitor do in a power
supply. There's also some resistances/capacitances in the AVC line.

But I could be wrong. If so, let me know!

I don't anything wrong with what you wrote but you seem to think that
the diode used makes no difference because you can make it up its
deficiencies with an amplifier whose input impedance and gain adjusts
for it. Basically that is true that you can use a less efficient diode
but you will have to provide higher signal levels to it and weak
signals will still be distorted due to compression. I suppose you could
use a logarithmic type amplifier following the detector in order to make
up for the compression.

If you look at the diode curves germanium has one of the better forward
current to input voltage ratios of several diode types. Not being a
radio designer my approach would be to use a diodes fairly liner region
with a better forward current to input voltage ratio where the least
distortion and compression would be due to it and therefor the least
needed correction to be made up for by a amplifier with a fixed
correction. Another reason to use a more efficient diode besides the
signal level power needed is the power the diode itself burns when you
bias the diodes with larger forward voltage junctions.

--
Telamon
Ventura, California

"Telamon" wrote in message
...

I don't anything wrong with what you wrote but you seem to think that
the diode used makes no difference because you can make it up its
deficiencies with an amplifier whose input impedance and gain adjusts
for it.

I don't think we disagree on anything important, but I wanted to say that,
after a point, it won't make any practical difference to the distortion of
the detector if a diode has a linear region or a very non linear square law
region. The resistance of the load soon dominates the characteristiscs of
the circuit.

The rest of my reply was mostly aimed at the original article's contention
that a diodes distortion level can be derived from only from a diode's
presumed square law characteristics and the modulation index.

Basically that is true that you can use a less efficient diode
but you will have to provide higher signal levels to it and weak
signals will still be distorted due to compression. I suppose you could
use a logarithmic type amplifier following the detector in order to make
up for the compression.

I suppose, but I don't see any need. The distortion of the diode detector
can be quite low if it's driven at a proper level to minimize the the amount
of the waveform in the non linear region of the detector.


If you look at the diode curves germanium has one of the better forward
current to input voltage ratios of several diode types.

Right. A germanium diode would generally give less distortion and better
sensitivity than a silicon diode. More than that, there used to be a bunch
of specialized germanium diodes intended for radio audio detection, video
detection and such. It seems now it's 1N34A types.


Not being a
radio designer my approach would be to use a diodes fairly liner region
with a better forward current to input voltage ratio where the least
distortion and compression would be due to it and therefor the least
needed correction to be made up for by a amplifier with a fixed
correction. Another reason to use a more efficient diode besides the
signal level power needed is the power the diode itself burns when you
bias the diodes with larger forward voltage junctions.


Efficiency is a bigger consideration with crystal sets.

Frank Dresser