Low-distortion AM detection?
Ross Archer
There are a number of really expert folks out here who might
know the answer to this question, so I'm posting here. It's
at least peripherally SWBC related, since SWBC is almost
always AM-modulated. :)
Did a fair amount of Googling on this issue, with no real
luck.
It emerges that the conventional diode "envelope" detector
which is used in virtually every AM-mode radio, and pretty
much any non-linear or square-law device, is going to
generate (along with DC and RF components and the desired
sudio modulation) a significant harmonic distortion
component. Second-harmonics especially, which I think are
often blamed for dissonant and unpleasant sound vs.
odd-order harmonics? Since the audio spectrum covers many
octaves, it's obvious that no low-pass filtering after the
fact will be able to eliminate all the harmonic content that
nonetheless falls within the audio passband.
It almost sounds as if this distortion is unavoidable with
non-linear devices (and of course, linear device can't
demodulate). So are we "stuck" with the distortion as the
price we pay for the simplicity of AM?
I was thinking that a diode forward-biased very close to
conduction might at least increase the S/N ratio of weak
signals by reducing the threshold effect on the recovered
half of the wave, but whether this would reduce distortion
is beyond me to calculate. The other thought would be to do
the detection after really vigorous AGC to keep the signal
swing small into the detector.
I've seen balanced mixers and amplifiers that "cancel" the
second-order harmonic, but not I'm sure if this has any
application for DSB with full carrier. Direct conversion
might be useful?
Can you get low distortion (meaning, around THD at 1% or
better?) from an AM detector from anything up to a 100%
modulated signal, or is that hopeless?
Thanks for any thoughts out there!
-- Ross
Maximo Lachman
> It almost sounds as if this distortion is unavoidable with
> non-linear devices (and of course, linear device can't
> demodulate). So are we "stuck" with the distortion as the
> price we pay for the simplicity of AM?
Did you find anything on full-wave detectors, which shows that they
reduce 2nd order harmonics? At least they have noticeably less
distortion than regular diode detectors.
> I was thinking that a diode forward-biased very close to
> conduction might at least increase the S/N ratio of weak
> signals by reducing the threshold effect on the recovered
> half of the wave, but whether this would reduce distortion
> is beyond me to calculate. The other thought would be to do
> the detection after really vigorous AGC to keep the signal
> swing small into the detector.
Full-wave detection is probably better, if not simpler.
> I've seen balanced mixers and amplifiers that "cancel" the
> second-order harmonic, but not I'm sure if this has any
> application for DSB with full carrier. Direct conversion
> might be useful?
Wouldn't direct conversion qualify as 'linear'? Anyway, it should do an
even better job than sync, according to reviews of the Liniplex.
> Can you get low distortion (meaning, around THD at 1% or
> better?) from an AM detector from anything up to a 100%
> modulated signal, or is that hopeless?
I posted a quote recently on the distortion introduced by sync detection,
and it was 'well under 1%'.
Ross Archer
news:arpo5t$1pq$1...@freenet9.carleton.ca:
> Ross Archer (arc...@topnow.com) writes:
>> It almost sounds as if this distortion is unavoidable with
>> non-linear devices (and of course, linear device can't
>> demodulate). So are we "stuck" with the distortion as the
>> price we pay for the simplicity of AM?
>
> Did you find anything on full-wave detectors, which shows that they
> reduce 2nd order harmonics? At least they have noticeably less
> distortion than regular diode detectors.
It's funny you mentioned that, as I recommended a full-wave circuit to
someone on here. But had no idea there might be distortion
ramifications.
>
>> I was thinking that a diode forward-biased very close to
>> conduction might at least increase the S/N ratio of weak
>> signals by reducing the threshold effect on the recovered
>> half of the wave, but whether this would reduce distortion
>> is beyond me to calculate. The other thought would be to do
>> the detection after really vigorous AGC to keep the signal
>> swing small into the detector.
>
> Full-wave detection is probably better, if not simpler.
Agreed. And simple is good for me. I make a lot of little circuits,
but seldom more than a few transistors or ICs total.
>
>> I've seen balanced mixers and amplifiers that "cancel" the
>> second-order harmonic, but not I'm sure if this has any
>> application for DSB with full carrier. Direct conversion
>> might be useful?
>
> Wouldn't direct conversion qualify as 'linear'? Anyway, it should do
> an even better job than sync, according to reviews of the Liniplex.
Can you provide a reference for Liniplex?
At any rate, there's no way a mixer could be linear. If it were, there
would be no mixing products. Just copies of the inputs.
>
>> Can you get low distortion (meaning, around THD at 1% or
>> better?) from an AM detector from anything up to a 100%
>> modulated signal, or is that hopeless?
>
> I posted a quote recently on the distortion introduced by sync
> detection, and it was 'well under 1%'.
Sync. detection is more complex that I want -- if there's a choice.
Granted, I really like the results of a good sync detector. :)
-- ross
Frank Dresser
>
>
>
>There are a number of really expert folks out here who might
>know the answer to this question, so I'm posting here. It's
>at least peripherally SWBC related, since SWBC is almost
>always AM-modulated. :)
>
>Did a fair amount of Googling on this issue, with no real
>luck.
>
>It emerges that the conventional diode "envelope" detector
>which is used in virtually every AM-mode radio, and pretty
>much any non-linear or square-law device, is going to
>generate (along with DC and RF components and the desired
>sudio modulation) a significant harmonic distortion
>component. Second-harmonics especially, which I think are
>often blamed for dissonant and unpleasant sound vs.
>odd-order harmonics? Since the audio spectrum covers many
>octaves, it's obvious that no low-pass filtering after the
>fact will be able to eliminate all the harmonic content that
>nonetheless falls within the audio passband.
>
>It almost sounds as if this distortion is unavoidable with
>non-linear devices (and of course, linear device can't
>demodulate). So are we "stuck" with the distortion as the
>price we pay for the simplicity of AM?
>
Think of the diode detector as a switch. It switches one half of the RF (or
IF) cycle on, and the other half off. A perfect switch would perfectly
reproduce the transmitted audio. Any flaws in the switching action would cause
distortion. Practical diode detectors are actually pretty good. The voltage -
current relationship worsens at very low voltages.
>I was thinking that a diode forward-biased very close to
>conduction might at least increase the S/N ratio of weak
>signals by reducing the threshold effect on the recovered
>half of the wave, but whether this would reduce distortion
>is beyond me to calculate. The other thought would be to do
>the detection after really vigorous AGC to keep the signal
>swing small into the detector.
>
Diode distortion is highest at full modulation. The negative peaks will get
clipped. A fully modulated 1V peak to peak signal will have much more of the
envelope in the higher distortion area than a 10 V peak to peak signal.
Signals at less than full modulation don't get close to the zero voltage point
and don't get distorted much by conventional diode detectors.
Ben Tongue, of Blonder-Tongue fame, has done some interesting investigations of
the voltage - current relationship of germanium diodes. This is not the more
familiar rigid "approximately .2V forward drop" rule:
http://uweb.superlink.net/bhtongue/7diodeCv/7diodeCv.html
A precision rectifier might be an alternative to a foreward biased diode:
http://www.play-hookey.com/analog/half-wave_rectifier.html
>I've seen balanced mixers and amplifiers that "cancel" the
>second-order harmonic, but not I'm sure if this has any
>application for DSB with full carrier. Direct conversion
>might be useful?
>
>Can you get low distortion (meaning, around THD at 1% or
>better?) from an AM detector from anything up to a 100%
>modulated signal, or is that hopeless?
Back in the 30's, some manufacturers were looking into very high level
detection schemes. The plan was to drive a high impedance loudspeaker directly
from a diode detector-filter. All the amplifacation was to be done at the
RF-IF stages. If you assume a carrier level of around maybe 50 volts and the
detector distortion came at under 1/2 volt, the total detector distotion would
be vanishingly small. But it was too expensive to be practical, and I doubt it
sounded any better. Detector distortion is only part of the total distortion,
and diode detectors work pretty well. Good enough for nearly all FM radios, as
well.
>
>Thanks for any thoughts out there!
>
>-- Ross
>
>
Frank Dresser
Brenda Ann
Actually, even order harmonics are much more pleasant, and much lower in
energy, than odd order harmonics. I.E. third harmonic is higher energy than
second, etc., filtering notwithstanding.
Also, full wave rectification is very rich in second order harmonics, at
least of the carrier frequency. For instance, a 60Hz sinewave, full wave
rectified, produces a 120Hz sawtooth waveform.
"Ross Archer" <arc...@topnow.com> wrote in message
news:3DE05C04...@topnow.com...
Maximo Lachman
>> reduce 2nd order harmonics? At least they have noticeably less
>> distortion than regular diode detectors.
>
> It's funny you mentioned that, as I recommended a full-wave circuit to
> someone on here. But had no idea there might be distortion
> ramifications.
It seems to help Grundig/Panasonic radios sound better than comparable
Sony/Sangean ones in AM mode.
>> Wouldn't direct conversion qualify as 'linear'? Anyway, it should do
>> an even better job than sync, according to reviews of the Liniplex.
>
> Can you provide a reference for Liniplex?
WRTH from the mid-80's, perhaps 1985. Otherwise, the inventor's patents
are online at the British Patent Office.
> At any rate, there's no way a mixer could be linear. If it were, there
> would be no mixing products. Just copies of the inputs.
By linear I mean the response curve, if the 2nd order errors are neglible
below 10 KHz.
Pete KE9OA
be interesting to do. Anyway, an SSB/CW detector is a mixer, except that at
the I.F. output, the sidebands are at audio vs. radio frequencies.
Pete
Pete KE9OA
been using both of these chips over the years. Another approach is to use
an op-amp with diodes in the feedback network, to work as a precision
full-wave detector. This will also give you low distortion. The idea of
slightly biasing a diode would also work ok. Another approach is shown in
the ARRL Handbook; this is described as an Infinite Impedance Detector,
using a JFET. I never did get that one to work.
Pete
Jake Brodsky
wrote:
>It almost sounds as if this distortion is unavoidable with
>non-linear devices (and of course, linear device can't
>demodulate). So are we "stuck" with the distortion as the
>price we pay for the simplicity of AM?
If you want to keep things stone simple the answer is Yes.
However...
>I was thinking that a diode forward-biased very close to
>conduction might at least increase the S/N ratio of weak
>signals by reducing the threshold effect on the recovered
>half of the wave, but whether this would reduce distortion
>is beyond me to calculate. The other thought would be to do
>the detection after really vigorous AGC to keep the signal
>swing small into the detector.
That's one way to handle it. Another way is to bias a single
transistor right at the ragged edge of conduction. With appropriate
RF choking, you might be able to recover a decent signal. Still, I
wouldn't expect much improvement from this method over a diode
alternative.
>I've seen balanced mixers and amplifiers that "cancel" the
>second-order harmonic, but not I'm sure if this has any
>application for DSB with full carrier. Direct conversion
>might be useful?
Motorola actually made some AM stereo demodulation chips with around
0.3 % THD. you might want to see what's available in that regard.
>Can you get low distortion (meaning, around THD at 1% or
>better?) from an AM detector from anything up to a 100%
>modulated signal, or is that hopeless?
It's been done.
>Thanks for any thoughts out there!
Jake Brodsky, AB3A mailto:fru...@erols.com
"Beware of the massive impossible!"
Tom2000
<n.giana...@worldnet.att.net> wrote:
>...Another approach is to use
>an op-amp with diodes in the feedback network, to work as a precision
>full-wave detector...
Pete,
I tried this years ago (back in the 70's) using, I think, 741's.
Those op amps, wired as precision rectifiers, didn't have nearly
enough slew speed to work behind a 455 kHz IF. The results were
pretty bad. (I don't remember if it didn't work at all, or if the
performance was just bad. I just remember that I abandoned the idea.)
Have you had good results with modern op amps?
Very 73,
Tom AB9B
DougVL
Chris in the U.K. He said:
"Found a good article in "Electronics World" ( formerly Wireless World )
on Homodyne and Synchrodyne AM receivers
"Synchrodyne/Homodyne Receiver"
Micheal Slifkin and Noam Dori
Electonics World November 1988 Volume 104 No. 1751 pp947-953
Electronics World article in pdf format is now attached (500k). Covers
design
of complete homodyne and synchrodyne receivers. Some of the Motorola
chips are now difficult to find but I guess substitutes could be found.
Here is the article: Synchr~1.pdf (516KB).
http://users.netonecom.net/~swordman/Radio/graphics/SYNCHR~1.PDF
Or you can find it on my "Short Wave Radio Page" at
http://users.netonecom.net/~swordman/Radio/ShortWaveRadio.htm
The authors claim that "The audio quality is similar to that of frequency
modulation." in relation to the synchrodyne. For the homodyne, they said
"The audio quality is certainly superior to that of a superheterodyne with
diode detection." It uses an MC1496 balanced modulator mixer for the
detector. One of their major design goals was an inexpensive system,
although they wanted to (and did) combine both systems into one.
--
DougVL, K8RFT
http://users.netonecom.net/~swordman/GetThePoint.htm
"In theory there's no difference between theory
and practice, but in practice...there is"
"Try to make it real....compared to what?"
"Ross Archer" <arc...@topnow.com> wrote in message
news:3DE05C04...@topnow.com...
>
> There are a number of really expert folks out here who might
> know the answer to this question, so I'm posting here. It's
> at least peripherally SWBC related, since SWBC is almost
> always AM-modulated. :)
>
> Did a fair amount of Googling on this issue, with no real
> luck.
>
snip
Ross Archer
>
> Either the TDA1072 or the TDA1572 provide less than .5% distortion.
Thanks for the pointers. I couldn't find the data-sheet for
1072, but pulled up the TDA1572, and it sounds like a
possible "build-it-yourself" project one day.
I note the data-sheet talks about full-wave balanced
envelope detector. Since it's on-chip and there are no
external leads from the detector to connect a transformer
to, I assume it's just a bridge made from carefully-matched
diodes.
I have
> been using both of these chips over the years. Another approach is to use
> an op-amp with diodes in the feedback network, to work as a precision
> full-wave detector. This will also give you low distortion. The idea of
> slightly biasing a diode would also work ok.
> Another approach is shown in
> the ARRL Handbook; this is described as an Infinite Impedance Detector,
> using a JFET. I never did get that one to work.
Hmm. Yeah, I suppose you have to tweak the gate bias to
that particular JFET's pinch-off voltage to make it work
just so. That would differ a lot from part to part even
from the same part number and batch?
>
> Pete
Thanks a lot for some ideas!
-- Ross
Ross Archer
>
> Ross Archer (dog...@mindless.com) writes:
> > di...@FreeNet.Carleton.CA (Maximo Lachman) wrote in
> >> Did you find anything on full-wave detectors, which shows that they
> >> reduce 2nd order harmonics? At least they have noticeably less
> >> distortion than regular diode detectors.
> >
> > It's funny you mentioned that, as I recommended a full-wave circuit to
> > someone on here. But had no idea there might be distortion
> > ramifications.
>
> It seems to help Grundig/Panasonic radios sound better than comparable
> Sony/Sangean ones in AM mode.
>
> >> Wouldn't direct conversion qualify as 'linear'? Anyway, it should do
> >> an even better job than sync, according to reviews of the Liniplex.
> >
> > Can you provide a reference for Liniplex?
>
> WRTH from the mid-80's, perhaps 1985. Otherwise, the inventor's patents
> are online at the British Patent Office.
>
> > At any rate, there's no way a mixer could be linear. If it were, there
> > would be no mixing products. Just copies of the inputs.
>
> By linear I mean the response curve, if the 2nd order errors are neglible
> below 10 KHz.
Ah.
Ross Archer
>
> On Mon, 25 Nov 2002 13:13:27 GMT, "Pete KE9OA"
> <n.giana...@worldnet.att.net> wrote:
>
> >...Another approach is to use
> >an op-amp with diodes in the feedback network, to work as a precision
> >full-wave detector...
>
> Pete,
>
> I tried this years ago (back in the 70's) using, I think, 741's.
> Those op amps, wired as precision rectifiers, didn't have nearly
> enough slew speed to work behind a 455 kHz IF. The results were
> pretty bad. (I don't remember if it didn't work at all, or if the
> performance was just bad. I just remember that I abandoned the idea.)
> Have you had good results with modern op amps?
>
> Very 73,
What is the concept behind a "precision rectifier"? Is it
to do away with the "step function"-like drop in current vs.
applied voltage when you get down to the junction
voltage-drop range?
Or is it doing something else?
I would think FETs (square law, so slope is linear at 2)
would work better than a diode (exponential at all
derivatives), so a FET biased at pinch-off might do the same
job with less complexity. My math is pretty weak here
though.
Ross Archer
Hmm. It still sounds like a plain-old "conduction angle"
issue? A class-B amplifier with a tiny bit of forward bias
should perform quite well then?
>
> >I was thinking that a diode forward-biased very close to
> >conduction might at least increase the S/N ratio of weak
> >signals by reducing the threshold effect on the recovered
> >half of the wave, but whether this would reduce distortion
> >is beyond me to calculate. The other thought would be to do
> >the detection after really vigorous AGC to keep the signal
> >swing small into the detector.
> >
>
> Diode distortion is highest at full modulation. The negative peaks will get
> clipped. A fully modulated 1V peak to peak signal will have much more of the
> envelope in the higher distortion area than a 10 V peak to peak signal.
> Signals at less than full modulation don't get close to the zero voltage point
> and don't get distorted much by conventional diode detectors.
>
> Ben Tongue, of Blonder-Tongue fame, has done some interesting investigations of
> the voltage - current relationship of germanium diodes. This is not the more
> familiar rigid "approximately .2V forward drop" rule:
>
> http://uweb.superlink.net/bhtongue/7diodeCv/7diodeCv.html
>
> A precision rectifier might be an alternative to a foreward biased diode:
>
> http://www.play-hookey.com/analog/half-wave_rectifier.html
Cool! The half-wave version looks simple enough and could
be made to work well enough with a single-ended supply.
>
> >I've seen balanced mixers and amplifiers that "cancel" the
> >second-order harmonic, but not I'm sure if this has any
> >application for DSB with full carrier. Direct conversion
> >might be useful?
> >
> >Can you get low distortion (meaning, around THD at 1% or
> >better?) from an AM detector from anything up to a 100%
> >modulated signal, or is that hopeless?
>
> Back in the 30's, some manufacturers were looking into very high level
> detection schemes. The plan was to drive a high impedance loudspeaker directly
> from a diode detector-filter. All the amplifacation was to be done at the
> RF-IF stages. If you assume a carrier level of around maybe 50 volts and the
> detector distortion came at under 1/2 volt, the total detector distotion would
> be vanishingly small.
1% of the time in distortion area so < 1% I guess. :)
> But it was too expensive to be practical, and I doubt it
> sounded any better. Detector distortion is only part of the total distortion,
> and diode detectors work pretty well. Good enough for nearly all FM radios, as
> well.
Probably right. Wouldn't want to know how much a typical
speaker adds. :)
THANKS. More cool links for my tinker-notebook!
-- Ross
Ross Archer
>
> Ross,
>
> Actually, even order harmonics are much more pleasant, and much lower in
> energy, than odd order harmonics. I.E. third harmonic is higher energy than
> second, etc., filtering notwithstanding.
I keep confusing those. Yup, what I said was backwards. :)
>
> Also, full wave rectification is very rich in second order harmonics, at
> least of the carrier frequency. For instance, a 60Hz sinewave, full wave
> rectified, produces a 120Hz sawtooth waveform.
I think that's no problem, because that would be the second
harmonic of the carrier frequency, not the modulation
frequency. This 2*carrier component would drop out in the
low-pass envelope filter, along with the original carrier
frequency and all the RF mixing components. All you'd get
is a DC component (probably twice as strong), which is
blocked by a series output capacitor, and the recovered
audio and whatever harmonics map into the audio response of
the audio stage.
Since the audio modulation is so far below the RF part in
frequency, the 2*carrier frequency part probably would have
to mix with very high-order audio harmonics to create any
mixing products that would fold back into the audio range.
I would imagine the added distortion would be negligible,
but that's a hand-waving argument. :)
Ross Archer
>
> On Sun, 24 Nov 2002 04:56:36 GMT, Ross Archer <arc...@topnow.com>
> wrote:
>
> >It almost sounds as if this distortion is unavoidable with
> >non-linear devices (and of course, linear device can't
> >demodulate). So are we "stuck" with the distortion as the
> >price we pay for the simplicity of AM?
>
> If you want to keep things stone simple the answer is Yes.
>
> However...
>
> >I was thinking that a diode forward-biased very close to
> >conduction might at least increase the S/N ratio of weak
> >signals by reducing the threshold effect on the recovered
> >half of the wave, but whether this would reduce distortion
> >is beyond me to calculate. The other thought would be to do
> >the detection after really vigorous AGC to keep the signal
> >swing small into the detector.
>
> That's one way to handle it. Another way is to bias a single
> transistor right at the ragged edge of conduction.
FET or bipolar?
> With appropriate
> RF choking, you might be able to recover a decent signal.
What would the RF choke do?
> Still, I
> wouldn't expect much improvement from this method over a diode
> alternative.
>
> >I've seen balanced mixers and amplifiers that "cancel" the
> >second-order harmonic, but not I'm sure if this has any
> >application for DSB with full carrier. Direct conversion
> >might be useful?
>
> Motorola actually made some AM stereo demodulation chips with around
> 0.3 % THD. you might want to see what's available in that regard.
I'll do that. Maybe one of them can be made to do sync
detector service, which would be even better I'd imagine.
>
> >Can you get low distortion (meaning, around THD at 1% or
> >better?) from an AM detector from anything up to a 100%
> >modulated signal, or is that hopeless?
>
> It's been done.
And a working model's the best possible argument for
something being possible.
"A thinking machine with judgement and intelligence that can
learn -- that's impossible!" -- Well, what are we? :)
Thanks for the ideas.
> >Thanks for any thoughts out there!
>
> Jake Brodsky, AB3A mailto:fru...@erols.com
> "Beware of the massive impossible!"
The moderately impossible can be a bear too.
John Doty
<arc...@topnow.com> wrote:
>> Another approach is shown in the ARRL Handbook; this is described as an
>> Infinite Impedance Detector, using a JFET. I never did get that one to
>> work.
> Hmm. Yeah, I suppose you have to tweak the gate bias to that particular
> JFET's pinch-off voltage to make it work just so. That would differ a
> lot from part to part even from the same part number and batch?
You just limit the DC source current to a low value (source
resistor>>/1/Gm), and it finds the right operating point automagically.
--
| John Doty "You can't confuse me, that's my job."
| Home: j...@w-d.org
| Work: j...@space.mit.edu
John Doty
<arc...@topnow.com> wrote:
> What is the concept behind a "precision rectifier"? Is it to do away
> with the "step function"-like drop in current vs. applied voltage when
> you get down to the junction voltage-drop range?
>
> Or is it doing something else?
There are a bunch of ways. Drake's envelope detector for the R8 series
runs the IF through a hard limiter, and then feeds the limited signal to
one input of a product detector, and the linear IF signal to the other
input, effectively computing sign(x)*x. Horowitz and Hill have an op
amp/diode scheme in their textbook.
The real distiction in envelope detectors is between peak detectors and
averaging detectors. "Infinite impedance" detectors and most diode
detectors are peak detectors. The Drake R8 detector is an averaging
detector. Either is theoretically capable of good performance, but they
work very differently.
>
> I would think FETs (square law, so slope is linear at 2) would work
> better than a diode (exponential at all derivatives), so a FET biased at
> pinch-off might do the same job with less complexity.
Near pinch-off, FETs have exponential current characteristics similar to
bipolars. The reason is that conduction in both cases is via thermal
electrons over a potential barrier. In the "square law" regime of FETs,
the electrons don't need a thermal kick to get through the barrier.
A square law is no good in any case. It not only mixes the carrier with
the sidebands, it mixes the sidebands with themselves and each other,
yielding lots of second-order audio distortion. The magic of a true
envelope detector is that the high order distortion cancels the lower
order distortion.
This cancellation depends on the phase/amplitude balance of the AM signal.
If you want good results from an envelope detector, you need to preceed it
with an IF with a symmetrical response and a constant group delay ("phase
linearity"). Avoiding multipath would also be nice (but harder to arrange
;-).
Synchronous detectors avoid this problem by separating the carrier from
the sidebands before detection, so only the desired products are present
at the detector output.
Jake Brodsky
>> That's one way to handle it. Another way is to bias a single
>> transistor right at the ragged edge of conduction.
>
>FET or bipolar?
You could do it with either one, but a Bipolar would probably be more
linear, once it gets in to conduction.
>> With appropriate
>> RF choking, you might be able to recover a decent signal.
>
>What would the RF choke do?
Well, you have an IF feeding in to a transistor amplifier. The
amplifier rectifies the signal, but you'll still have quite a bit of
an RF component to the output. You probably should shunt a bypass to
ground and choke the RF from the audio before coupling it to a
following audio stage.
>> Motorola actually made some AM stereo demodulation chips with around
>> 0.3 % THD. you might want to see what's available in that regard.
>
>I'll do that. Maybe one of them can be made to do sync
>detector service, which would be even better I'd imagine.
The R-75 uses one. It's Motorola's MC13020 series Stereo chip. I
found a similar version of this chip documentation here:
http://merchant.hibbertco.com/fs22/deact/fs22/pdf-docs/motorola/mc13022a.rev1.pdf
It boasts a typical THD of 0.3% for monaural reception of a signal
with 50% modulation. Worst case THD is 1.6% for Stereo reception,
typical THD for Stereo is 0.8%.
>> "Beware of the massive impossible!"
>
>The moderately impossible can be a bear too.
This was just one of many brain droppings I've had. It wasn't really
meant to be deep, though many who read it think it is :-)
Frank Dresser
>> Think of the diode detector as a switch. It switches one half of the RF
>(or
>> IF) cycle on, and the other half off. A perfect switch would perfectly
>> reproduce the transmitted audio. Any flaws in the switching action would
>cause
>> distortion. Practical diode detectors are actually pretty good. The
>voltage -
>> current relationship worsens at very low voltages.
>
>Hmm. It still sounds like a plain-old "conduction angle"
>issue? A class-B amplifier with a tiny bit of forward bias
>should perform quite well then?
>>
That sounds like the old plate detector from the vacuum tube days. It was
reasonably sensitive, but had increased distortion due to the nonlinear changes
in amplification near cutoff. This is similiar to the problems diodes have at
low voltages, but you can't reduce distortion in a plate detector with higher
drive as with a diode, because tube saturation causes distortion on the high
drive end, as well. I'd expect transistors to act about the same.
>> A precision rectifier might be an alternative to a foreward biased diode:
>>
>> http://www.play-hookey.com/analog/half-wave_rectifier.html
>
>Cool! The half-wave version looks simple enough and could
>be made to work well enough with a single-ended supply.
You've got to watch the slew rate at RF/IF frequencies. 0 to peak at 500khz is
.5microsecond. A 741 has a slew rate of .5V/us. A LM6361 has a slew rate of
300V/us. At least that's what my catalog says. Some op amps have a small dead
band right around 0v. Some are kinda noisy. Anyway, it's reasonably simple.
An even simpler thing is making a diode by connecting the base and collector of
a transistor together. Germanium would be best for a detector. A little bit
of base current will multiply itself through the collector. It's sort of an
amplified diode. Doesn't really fix the troublesome very low voltage areas and
it has increased capacitance, though.
>> >
>> >Can you get low distortion (meaning, around THD at 1% or
>> >better?) from an AM detector from anything up to a 100%
>> >modulated signal, or is that hopeless?
>>
>> Back in the 30's, some manufacturers were looking into very high level
>> detection schemes. The plan was to drive a high impedance loudspeaker
>directly
>> from a diode detector-filter. All the amplifacation was to be done at the
>> RF-IF stages. If you assume a carrier level of around maybe 50 volts and
>the
>> detector distortion came at under 1/2 volt, the total detector distotion
>would
>> be vanishingly small.
>
>1% of the time in distortion area so < 1% I guess. :)
>
To quote the Radiotron Designer's Handbook (third edition) on detectors:
"Summing up the characteristics of the diode, it may be stated that its
performance, as regards harmonic distortion and frequency response, is
excellent provided that the input voltage is sufficently high and that A.C.
shunting is reduced to a minimum. All forms of detectors suffer from
distortion at low input levels, but the diode has the particular advantage that
the input may be increased to a very high level with consequent feduction of
distortion without any overloading effect such as occurs with other forms of
detectors."
By the way, the ARRL handbooks have good general descriptions of various
detector circuits. Check your library, you might find one that goes way back.
>> But it was too expensive to be practical, and I doubt it
>> sounded any better. Detector distortion is only part of the total
>distortion,
>> and diode detectors work pretty well. Good enough for nearly all FM
>radios, as
>> well.
>
>Probably right. Wouldn't want to know how much a typical
>speaker adds. :)
>
Mixer, variable gain stages, hum, noise from everything. Well, if you don't
hear it, you don't worry about it.
>THANKS. More cool links for my tinker-notebook!
>
OK, here's a sync detector link:
http://home.att.net/~wa1sov/technical/sync_det.html
>-- Ross
>
Hey, if you come up with anything interesting, please post it.
Frank Dresser
Michael Black
> In August, 2002 I received a scan of a synchronous detection article from
> Chris in the U.K. He said:
>
> "Found a good article in "Electronics World" ( formerly Wireless World )
> on Homodyne and Synchrodyne AM receivers
>
> "Synchrodyne/Homodyne Receiver"
> Micheal Slifkin and Noam Dori
> Electonics World November 1988 Volume 104 No. 1751 pp947-953
>
I was thinking about a lot easier than I expected.
In the January 1986 issue of "Electronics & Wireless World" starts
a three-part article by Linsely Hood, "Synchrodyne A.M. Receiver".
It uses MC1496 double-balanced mixers as phase detector and product
detector.
Michael
Michael Black
> On Mon, 25 Nov 2002 13:13:27 GMT, "Pete KE9OA"
> <n.giana...@worldnet.att.net> wrote:
>
> >...Another approach is to use
> >an op-amp with diodes in the feedback network, to work as a precision
> >full-wave detector...
>
> Pete,
>
> I tried this years ago (back in the 70's) using, I think, 741's.
> Those op amps, wired as precision rectifiers, didn't have nearly
> enough slew speed to work behind a 455 kHz IF. The results were
> pretty bad. (I don't remember if it didn't work at all, or if the
> performance was just bad. I just remember that I abandoned the idea.)
> Have you had good results with modern op amps?
>
> Very 73,
>
> Tom AB9B
>
the common op-amps were fairly limited for even low-RF use.
In the January 1976 issue of Ham Radio, Hank Olson has an article
entitled "Diode Detectors" which I remember as a comprehensive
article about AM detectors (around the same time there was an article
about FM detectors that covered just about any specific type). But really,
it starts out with catwhiskers through germanium and silicon diodes (not
really covering tube diodes, puts in a bit about synchronous detectors
(which were being covered pretty well at that time, thanks to IC phase
locked-loops), and then to precision diode detectors. In other words,
it really seems to be an opportunity for him to discuss precision diode
detectors for AM detection.
He shows LM318s in the circuit, which were better than 741's and probably
some of the best readily available to hobbyists at the time, but not
as good as some later common and cheap op-amps.
Around that same time, or maybe a few years later, someone showed a simple
"crystal radio" using a precision rectifier, in one of those reader
submission columns in Electronics. It used a 741, and had little or
not gain, to improve it's bandwidth. It ran in the broadcast band, with
just a loopstick and variable capacitor ahead of it. The circuit
also showed up in a Forest Mims column in Popular Electronics. I remember
trying it, and not getting much out of it.
But around that time, there was a fair amount of work on active AM
detectors, presumably because linear ICs were just ramping up and people
were into trying new things.
The LM372 was an "AM IF strip" that used an active AM detector. There
was a good article about its internals in 73 magazine for November 1972,
a mere thirty years ago, by someone who worked at National, maybe
the designer. Oddly, I never saw much about the IC elsewhere, and
can't remember seeing it listed for sale in the usual hobby places.
Likewise, the National LM373 had an active AM detector. This was
a whole multi-mode IF strip, useable for FM, AM, and SSB. The
double balanced mixer was used for the quadrature detector, as
a product detector for SSB, and unbalanced for AM, so the IF
signal would feed through to the active detector that was otherwise
used to generate AGC voltage. Of course, one of the application notes
showed it used as a quasi-synchronous detector, using the balanced
mixer and feeding it with the IF and the IF that had been limited.
Somewhere I have an application note by one of the lesser IC manufacturers
about active AM detectors.
The Drake SPR-4 had an AM detector that seems to be just a forward biased
germanium (1N270) diode directly feeding the base of a bipolar transistor.
The circuit is shown in the product review in QST for December 1970. (I
think the same circuit showed up in some later ARRL Handbooks.)
A similar circuit, using a hot-carrier diode, is shown in a receiver
construction article in Ham Radio for July 1978.
Rohde in one of his articles in Ham Radio in the late seventies had
a circuit or two for a better than average AM detector, which I'm not
going to look for at the moment. It's not in his receiver book.
When the Drake R8 came out, the article in Communications Quarterly
about its design mentioned that while the R7 claimed a synchronous
detector, it wasn't. I don't have the article handy, but perhaps
the R7 used some sort of active detector.
Elektor in it's October 1991 issue (I assume it's the British version;
definitely the article is in English) had an article by R. Shankar,
"AM Broadcast Receiver" which was a fairly standard superhet, but
used some interesting circuits, including an active detector using
bipolar transistors.
I have a vague recollection of seeing a whole AM BCB receiver described
somewhere that used a precision rectifier (ie with op-amps), and assume
it was in Wireless World. But I can't find the folder where I'd expect
I filed it, and I'm not even sure if I'm confusing a memory with that
Elektor article.
Michael
MRe
"Michael Black" <blac...@cam.org> schreef in bericht
news:6447bcd3.02112...@posting.google.com...
Way back somewhere in 1964 a Dutch electronics magazine had
a project for a home built shortwave reciever with a SSSB
detector (Synchronised single
sideband) as they called it. It was fully equiped with
tubes.
Synchronisation with a (penthode)tube employed in a
reactance circuit which tuned the BFO. Reactance circuit
driven by the output of a FM-discrimminator circuit.
The autor of this article was very clear about the purpose
and the ins an outs of synchronous detection of AM Signals.
This was 1964!
Martijn R.
Maximo Lachman
> Maximo Lachman wrote:
>>
>> It seems to help Grundig/Panasonic radios sound better than comparable
>> Sony/Sangean ones in AM mode.
> Grundig and Panasonic routinely use a full-wave detector?
Yes.
>> > Can you provide a reference for Liniplex?
>>
>> WRTH from the mid-80's, perhaps 1985. Otherwise, the inventor's patents
>> are online at the British Patent Office.
Here's some more info on E.C. Forster's patent:
I/Q Asynchronous AM Receiver
(GB 2295513, 1994/1998)
* For implementation of single chip high performance AM receivers
* Simple accurate analogue computing technique
* I/Q direct conversion (homodyne) architecture
* Consistent and repeatable performance
which is a more modern reworking of the Liniplex that relied on
crystals rather than a chip.
Ross Archer
news:fxGdnTEB3Mc...@speakeasy.net:
That's very interesting. Thank you for clarifying this.
>
> This cancellation depends on the phase/amplitude balance of the AM
> signal. If you want good results from an envelope detector, you need
> to preceed it with an IF with a symmetrical response and a constant
> group delay ("phase linearity"). Avoiding multipath would also be nice
> (but harder to arrange ;-).
>
> Synchronous detectors avoid this problem by separating the carrier
> from the sidebands before detection, so only the desired products are
> present at the detector output.
>
I figure that some sort of sync. detector would probably be best, but
far from a casual project.
The half-wave precision rectifier arrangement looks promising if I can
find an op-amp fast enough to have good open-loop gain even at 1.7 Mhz
for a TRF receiver so it can have a substantial gain increase when the
feedback loop is opened vs. closed.
Ross Archer
news:3de2d9a9$0$194$45be...@newscene.com:
> In August, 2002 I received a scan of a synchronous detection article
> from Chris in the U.K. He said:
>
> "Found a good article in "Electronics World" ( formerly Wireless World
> ) on Homodyne and Synchrodyne AM receivers
>
> "Synchrodyne/Homodyne Receiver"
> Micheal Slifkin and Noam Dori
> Electonics World November 1988 Volume 104 No. 1751 pp947-953
>
> Electronics World article in pdf format is now attached (500k). Covers
> design
> of complete homodyne and synchrodyne receivers. Some of the Motorola
> chips are now difficult to find but I guess substitutes could be
> found.
>
> Here is the article: Synchr~1.pdf (516KB).
> http://users.netonecom.net/~swordman/Radio/graphics/SYNCHR~1.PDF
>
> Or you can find it on my "Short Wave Radio Page" at
> http://users.netonecom.net/~swordman/Radio/ShortWaveRadio.htm
Thanks a lot for this. It's very interesting. Never thought about the
hetrodyne problem before.
>
>
> The authors claim that "The audio quality is similar to that of
> frequency modulation."
I wouldn't be surprised. I recently listened to 610 AM KFRC (oldies
music) with the sync detector on the R8B (both sidebands) and wide
filter, with Tivoli radio as audio out. It sounded really great.
-- Ross
John Doty
<dog...@mindless.com> wrote:
> The half-wave precision rectifier arrangement looks promising if I can
> find an op-amp fast enough to have good open-loop gain even at 1.7 Mhz
> for a TRF receiver so it can have a substantial gain increase when the
> feedback loop is opened vs. closed.
Fast nonlinear op amp circuits are a bit tricky. For precise performance
at 1.7 MHz, you'll need an amplifier with a gain-bandwidth product of
hundreds of MHz. These are not difficult to find, but they are difficult
to use: a well designed printed circuit board is essential, and surface
mount components are desirable.
The Drake approach is much less touchy, as it involves no feedback. A
comparator (to act as the limiter) and a mixer (or even just a
voltage-controlled switch) should be relatively easy to set up. Drake
uses an AC-coupled CMOS logic gate biased to threshold as a comparator,
with a little positive feedback to insure clean switching. The output
goes to the 1496 mixer.