In article ,
writes:

Experimental Methods in RF Design points out that direct conversion
receivers have become highly popular in the past couple of decades... this
seems somewhat surprising; I would have guessed people back in the, e.g.,
'60s, would have gone to great lengths to avoid image reject filters and
long IF chains.

The nice thing about DC IQ receivers (apart from their zero image problem) is
that, any kind of demodulation can be solved in software, and is fully
updatable ... whereas if it's done in hardware, you'd need new hardware for
each mode
required etc.

The "first" complete receivers for HF were done over 30 years ago.
Totally in hardware.
Equally capable of CW or SSB reception without switching anything.

The lack of image problems, simplicity of hardware and fully updatable
modulation schemes is what makes DC IQ so nice -. so it's not surprising to
me at all why it's becoming so popular.

Demodulation for In-phase and Quadrature had a much wider
application than you would normally consider, and by the millions
without software of any kind: NTSC chrominance demodulation in
TV receivers. Did it at the HF level, too, 3.58 MHz. :-)

I'm not positive about it, but I believe the first "image-less" mixer
systems were for radar applications in the 1950s, specifically for
monopulse radar tracking. I first encountered monopulse around
1959.

I agree that most everything can be done in software...provided one
gets the proper A-to-D arrangement capable of operating at extremely
low noise levels. Such is NOT easy.

DC receivers in general have many good features. But, one has to
do a realistic comparison versus the superheterodyne structure.

1. To get good sensitivity one has to work with truly low-level
signals that cannot take advantage of narrow bandpass filtering
ahead of the demodulator. A DC mixer input is relatively
broadband and that increases the total noise power in the
circuit.

2. Relative broadbandedness of the input absolutely requires a
high IP3 since adjacent, normally-unheard signals can be at
a high relative level. Few DC receivers have any AGC to the
input stages.

3. The "image-less" condition (receiving only high-side or low-
side of LO frequency) is resolved in demodulated audio and
that absolutely requires a broadband phase-shifting network.
Even with high IP3 specs on the input mixer, strong input
signals adjacent in frequency can get through and fail to be
cancelled in the audio network...if the adjacent signals are
outside of the phasing network's range.

4. If the phasing network is simulated in software, then the
microprocessor or microcontroller must be adequately
shielded to avoid transitent RFI from getting into the input.
Physical proximity would be very close and the input has no
narrowband filtering to help that. Software demodulation
will depend heavily on the type of processor and a lot of
specs that have no direct relationship to software.

The end result is really a compromise of fewer parts traded for
a whole new set of potential problems. One kind is not
"superior" to another kind, just different.

Len Anderson
retired (from regular hours) electronic engineer person

Avery Fineman wrote:
Can one get separated sidebands on AM DSB with a DC receiver?
Absolutely!

That's good to know. At present I'm going to quit performing these
"intellectual experiments" and start building something, and while I'm after
C-QUAM AM stereo (rather than upper/lower sideband stereo), it's good to
know what else _could_ be received.

BTW, if anyone wants to see the block diagram of what I'm planning to do,
see he http://oregonstate.edu/~kolstadj/RadioProj.gif . Keep in mind
it's designed primarily for simplicity, not for phoenomenally good noise
performance, sensitivity, selectivity, etc.

(I'd be particularly interested in comments on how to implement the low pass
filters -- it seems one would want phase preserving filters such as Bessels
or a cascade of a Chebyshev followed by an all-pass phase restoration
filter.)

Nooo...AM "came about" with absurdly SIMPLE components first,
not even using any vacuum tubes!

Wow... I realize now there's a large gap in my knowledge of the history of
the progression of radio inbetween "spark gap transmitter" and "diode-based
envelope detector!" I have read of coherers before in Lee's book, "Design
of CMOS Radio-Frequency Integrated Circuits" where he claims that nobody
ever really did figure out _how_ they worked -- interest wanted as better
detectors were available before they were around long enough for someone to
do so.

You have a phoenomenal memory, Len... I wish I could recall the details as
well as you have!

Long-distance telephony was the birthplace of SSB. Frequency
multiplexing was the only practical way to cram four telephone
circuits on a single pair of wires running many miles way back when.

If you tell me people were already using IQ modulation back then as well
I'll be quite impressed...

If you want synchronous detection of AM DSB, then you concentrate
on getting a carrier reinsertion oscillator locked to the received
carrier. Primary object is to get that lock.

I'm planning to write a (software) quadrature detector, and once that works,
start worrying about obtaining phase lock so that stereo can be decoded.

Can you get a synchronous detection of AM SSB? Difficult unless
the transmitter at the other end has sloppy carrier suppression.

Without a carrier of some pilot tone (as a reference) it seems as though
it's difficult to even claim there could be such a thing as 'synchronous
detection.'

DC receivers (also called "Zero-IF") came into popularity in Europe
THREE decades ago. RSGB's Radio Communication magazines of
1973 were showing stuff in Pat Hawker's monthly column. I got
interested in the Mike Gingell polyphase R-C network by seeing it
first in there.

I took a quick look at the Gingell networks and they seem quite novel --
even made their way into a Real Commerical Product (a Maxim IC).
(Interestingly enough, Dr. Gabor Temes -- who spent a long time designing
telephone network filters before going into academia, where he is now, all
of about 500' away from me here -- says there is still some black magic
involved in making them work. :-) )

Thanks for all the advice Len... I'd be offering to take you to dinner by
now if you were halfway local!

---Joel

Avery Fineman wrote:
Can one get separated sidebands on AM DSB with a DC receiver?
Absolutely!

That's good to know. At present I'm going to quit performing these
"intellectual experiments" and start building something, and while I'm after
C-QUAM AM stereo (rather than upper/lower sideband stereo), it's good to
know what else _could_ be received.

BTW, if anyone wants to see the block diagram of what I'm planning to do,
see he http://oregonstate.edu/~kolstadj/RadioProj.gif . Keep in mind
it's designed primarily for simplicity, not for phoenomenally good noise
performance, sensitivity, selectivity, etc.

(I'd be particularly interested in comments on how to implement the low pass
filters -- it seems one would want phase preserving filters such as Bessels
or a cascade of a Chebyshev followed by an all-pass phase restoration
filter.)

Nooo...AM "came about" with absurdly SIMPLE components first,
not even using any vacuum tubes!

Wow... I realize now there's a large gap in my knowledge of the history of
the progression of radio inbetween "spark gap transmitter" and "diode-based
envelope detector!" I have read of coherers before in Lee's book, "Design
of CMOS Radio-Frequency Integrated Circuits" where he claims that nobody
ever really did figure out _how_ they worked -- interest wanted as better
detectors were available before they were around long enough for someone to
do so.

You have a phoenomenal memory, Len... I wish I could recall the details as
well as you have!

Long-distance telephony was the birthplace of SSB. Frequency
multiplexing was the only practical way to cram four telephone
circuits on a single pair of wires running many miles way back when.

If you tell me people were already using IQ modulation back then as well
I'll be quite impressed...

If you want synchronous detection of AM DSB, then you concentrate
on getting a carrier reinsertion oscillator locked to the received
carrier. Primary object is to get that lock.

I'm planning to write a (software) quadrature detector, and once that works,
start worrying about obtaining phase lock so that stereo can be decoded.

Can you get a synchronous detection of AM SSB? Difficult unless
the transmitter at the other end has sloppy carrier suppression.

Without a carrier of some pilot tone (as a reference) it seems as though
it's difficult to even claim there could be such a thing as 'synchronous
detection.'

DC receivers (also called "Zero-IF") came into popularity in Europe
THREE decades ago. RSGB's Radio Communication magazines of
1973 were showing stuff in Pat Hawker's monthly column. I got
interested in the Mike Gingell polyphase R-C network by seeing it
first in there.

I took a quick look at the Gingell networks and they seem quite novel --
even made their way into a Real Commerical Product (a Maxim IC).
(Interestingly enough, Dr. Gabor Temes -- who spent a long time designing
telephone network filters before going into academia, where he is now, all
of about 500' away from me here -- says there is still some black magic
involved in making them work. :-) )

Thanks for all the advice Len... I'd be offering to take you to dinner by
now if you were halfway local!

---Joel

Bill Meara wrote:
Joel: Image reject filters? Long IF chains? My DC recievers have
neither.

That's what I meant -- DC receivers don't need them, and therefore it's
surprising DC receivers have only really been gaining steam here in the U.S.
in the past decade or two.

Then again, there's no arguing with performance: The GE "super radio" is, I
believe, a triple conversion receiver!

---Joel Kolstad

Bill Meara wrote:
Joel: Image reject filters? Long IF chains? My DC recievers have
neither.

That's what I meant -- DC receivers don't need them, and therefore it's
surprising DC receivers have only really been gaining steam here in the U.S.
in the past decade or two.

Then again, there's no arguing with performance: The GE "super radio" is, I
believe, a triple conversion receiver!

---Joel Kolstad

In article , "Joel Kolstad"
writes:

Avery Fineman wrote:
Can one get separated sidebands on AM DSB with a DC receiver?
Absolutely!

That's good to know. At present I'm going to quit performing these
"intellectual experiments" and start building something, and while I'm after
C-QUAM AM stereo (rather than upper/lower sideband stereo), it's good to
know what else _could_ be received.

BTW, if anyone wants to see the block diagram of what I'm planning to do,
see he http://oregonstate.edu/~kolstadj/RadioProj.gif . Keep in mind
it's designed primarily for simplicity, not for phoenomenally good noise
performance, sensitivity, selectivity, etc.

(I'd be particularly interested in comments on how to implement the low pass
filters -- it seems one would want phase preserving filters such as Bessels
or a cascade of a Chebyshev followed by an all-pass phase restoration
filter.)

OK, I got the block diagram. If you are using even a rudimentary
R-C lowpass following the two mixers, you need the parts rather
well matched in order to preserve identical relative phases. It is
important to HOLD the relative phase error at audio to a very small
number in order to do the In-phase/Quadrature thing. DSP will
work with BOTH magnitude and phase regardless of the kind of
modulation going into the mixers. You CAN realize a lowpass
function in DSP but the TI chip inputs probably needs some sort
of hardware lowpass filtering...? A simple R-C lowpass can be
checked out independently just from equal parts values to assure
minimum relative phase error.

Here's a good hint on melding hardware with software using DSP:

"Scientist's and Engineer's Guide to Digital Signal Processing,"
by Stephen W. Smith, PhD, California Technical Publishing.

Despite the title, this is a good text on DSP from the beginner's
point of view on to the more advanced. What is special is that
ALL the chapters can be downloaded absolutely FREE! :-)
[or pay about $68 for the hardcover]

http://www.DSPguide.com

Well organized book and a good "teaching style" to the writing.

Once you have the hardware fairly well in shape, it's time to go
nuts with the programming. This book ought to help whatever it
is you are going to code.

Nooo...AM "came about" with absurdly SIMPLE components first,
not even using any vacuum tubes!

Wow... I realize now there's a large gap in my knowledge of the history of
the progression of radio inbetween "spark gap transmitter" and "diode-based
envelope detector!" I have read of coherers before in Lee's book, "Design
of CMOS Radio-Frequency Integrated Circuits" where he claims that nobody
ever really did figure out _how_ they worked -- interest wanted as better
detectors were available before they were around long enough for someone to
do so.

Actually, that's irrelevant and a historical curiosity.

The galena crystal and "cat's whisker" formed a rudimentary point-
contact diode. I had one of those in 1946, a Philmore Crystal Set my
Dad got for me (el cheapo quality, but it worked after a fashion). A
half year later a new electronics store opened up in town and they
were selling surplus WW2 radar set silicon mixer diodes, type 1N21
and 1N23. Put one of those in the Philmore and really "souped up"
the audio. :-)

Long-distance telephony was the birthplace of SSB. Frequency
multiplexing was the only practical way to cram four telephone
circuits on a single pair of wires running many miles way back when.

If you tell me people were already using IQ modulation back then as well
I'll be quite impressed...

As far as I've seen, the old telephony "carrier" equipment used ordinary
4-diode ring mixers, usually copper-oxide stacked plate types, the
small ones the size of old multimeter AC rectifiers. That was pre-1930.

I'm not sure when the In-phase/Quadrature demod/mod sub-systems
were first used other than probably just before 1940...or maybe in the
WW2 years. I know the beginning applications were there in the late
1940s.

If you want synchronous detection of AM DSB, then you concentrate
on getting a carrier reinsertion oscillator locked to the received
carrier. Primary object is to get that lock.

I'm planning to write a (software) quadrature detector, and once that works,
start worrying about obtaining phase lock so that stereo can be decoded.

Good luck on that.

Can you get a synchronous detection of AM SSB? Difficult unless
the transmitter at the other end has sloppy carrier suppression.

Without a carrier of some pilot tone (as a reference) it seems as though
it's difficult to even claim there could be such a thing as 'synchronous
detection.'

I've seen it claimed in text, but no details, that a quasi-lock could
be obtained via voice, working on the harmonics of speech tones.
I'm not going to buy that until I see a demo.

DC receivers (also called "Zero-IF") came into popularity in Europe
THREE decades ago. RSGB's Radio Communication magazines of
1973 were showing stuff in Pat Hawker's monthly column. I got
interested in the Mike Gingell polyphase R-C network by seeing it
first in there.

I took a quick look at the Gingell networks and they seem quite novel --
even made their way into a Real Commerical Product (a Maxim IC).
(Interestingly enough, Dr. Gabor Temes -- who spent a long time designing
telephone network filters before going into academia, where he is now, all
of about 500' away from me here -- says there is still some black magic
involved in making them work. :-) )

The polyphase network was the subject of Michael Gingell's PhD
thesis in the UK. Material on that was seen on the Internet. Mike
is a USA resident now (or was a couple years ago when he had a
website...has a US ham callsign, too). A Japanese ham got busy
on that polyphase network and came up with an optimum set of
component values. That was published in QEX. Hans Summers'
website has links to all those.

Gabor Temes is a familiar name to textbook thumbers. :-) There
isn't a lot of black magic associated with the Gingell network, but it is
a thorough #$%^!!!! to try and analyze with its busy interconnections.
I stole a few minutes of CPU time on the RCA corporate computer,
using their LECAP (a frequency-domain version of IBM's ECAP...the
SPICE thing hadn't been developed yet) back in the 1970s. It worked
as advertised with only 0 and 180 degree audio input, producing nice
relative quadrature phases on all four outputs. Was surprised!

Some scrounged parts, not well measured as to values, and a quick
open-air toss-together showed excellent broadband relative quadrature
with less than 1 degree error in the 'voice' bandspace. Checked that
with a time-interval function on a homebuilt frequency counter.

Thanks for all the advice Len... I'd be offering to take you to dinner by
now if you were halfway local!

Thank you but I'll just wave as my wife and I roll through Oregon
along I-5 about once a year from southern California to Puget Sound
area of Washington. :-) Want to bypass the usual clogging just
before crossing the river into WA and vice-versa.

Len Anderson
retired (from regular hours) electronic engineer person

In article , "Joel Kolstad"
writes:

Avery Fineman wrote:
Can one get separated sidebands on AM DSB with a DC receiver?
Absolutely!

That's good to know. At present I'm going to quit performing these
"intellectual experiments" and start building something, and while I'm after
C-QUAM AM stereo (rather than upper/lower sideband stereo), it's good to
know what else _could_ be received.

BTW, if anyone wants to see the block diagram of what I'm planning to do,
see he http://oregonstate.edu/~kolstadj/RadioProj.gif . Keep in mind
it's designed primarily for simplicity, not for phoenomenally good noise
performance, sensitivity, selectivity, etc.

(I'd be particularly interested in comments on how to implement the low pass
filters -- it seems one would want phase preserving filters such as Bessels
or a cascade of a Chebyshev followed by an all-pass phase restoration
filter.)

OK, I got the block diagram. If you are using even a rudimentary
R-C lowpass following the two mixers, you need the parts rather
well matched in order to preserve identical relative phases. It is
important to HOLD the relative phase error at audio to a very small
number in order to do the In-phase/Quadrature thing. DSP will
work with BOTH magnitude and phase regardless of the kind of
modulation going into the mixers. You CAN realize a lowpass
function in DSP but the TI chip inputs probably needs some sort
of hardware lowpass filtering...? A simple R-C lowpass can be
checked out independently just from equal parts values to assure
minimum relative phase error.

Here's a good hint on melding hardware with software using DSP:

"Scientist's and Engineer's Guide to Digital Signal Processing,"
by Stephen W. Smith, PhD, California Technical Publishing.

Despite the title, this is a good text on DSP from the beginner's
point of view on to the more advanced. What is special is that
ALL the chapters can be downloaded absolutely FREE! :-)
[or pay about $68 for the hardcover]

http://www.DSPguide.com

Well organized book and a good "teaching style" to the writing.

Once you have the hardware fairly well in shape, it's time to go
nuts with the programming. This book ought to help whatever it
is you are going to code.

Nooo...AM "came about" with absurdly SIMPLE components first,
not even using any vacuum tubes!

Wow... I realize now there's a large gap in my knowledge of the history of
the progression of radio inbetween "spark gap transmitter" and "diode-based
envelope detector!" I have read of coherers before in Lee's book, "Design
of CMOS Radio-Frequency Integrated Circuits" where he claims that nobody
ever really did figure out _how_ they worked -- interest wanted as better
detectors were available before they were around long enough for someone to
do so.

Actually, that's irrelevant and a historical curiosity.

The galena crystal and "cat's whisker" formed a rudimentary point-
contact diode. I had one of those in 1946, a Philmore Crystal Set my
Dad got for me (el cheapo quality, but it worked after a fashion). A
half year later a new electronics store opened up in town and they
were selling surplus WW2 radar set silicon mixer diodes, type 1N21
and 1N23. Put one of those in the Philmore and really "souped up"
the audio. :-)

Long-distance telephony was the birthplace of SSB. Frequency
multiplexing was the only practical way to cram four telephone
circuits on a single pair of wires running many miles way back when.

If you tell me people were already using IQ modulation back then as well
I'll be quite impressed...

As far as I've seen, the old telephony "carrier" equipment used ordinary
4-diode ring mixers, usually copper-oxide stacked plate types, the
small ones the size of old multimeter AC rectifiers. That was pre-1930.

I'm not sure when the In-phase/Quadrature demod/mod sub-systems
were first used other than probably just before 1940...or maybe in the
WW2 years. I know the beginning applications were there in the late
1940s.

If you want synchronous detection of AM DSB, then you concentrate
on getting a carrier reinsertion oscillator locked to the received
carrier. Primary object is to get that lock.

I'm planning to write a (software) quadrature detector, and once that works,
start worrying about obtaining phase lock so that stereo can be decoded.

Good luck on that.

Can you get a synchronous detection of AM SSB? Difficult unless
the transmitter at the other end has sloppy carrier suppression.

Without a carrier of some pilot tone (as a reference) it seems as though
it's difficult to even claim there could be such a thing as 'synchronous
detection.'

I've seen it claimed in text, but no details, that a quasi-lock could
be obtained via voice, working on the harmonics of speech tones.
I'm not going to buy that until I see a demo.

DC receivers (also called "Zero-IF") came into popularity in Europe
THREE decades ago. RSGB's Radio Communication magazines of
1973 were showing stuff in Pat Hawker's monthly column. I got
interested in the Mike Gingell polyphase R-C network by seeing it
first in there.

I took a quick look at the Gingell networks and they seem quite novel --
even made their way into a Real Commerical Product (a Maxim IC).
(Interestingly enough, Dr. Gabor Temes -- who spent a long time designing
telephone network filters before going into academia, where he is now, all
of about 500' away from me here -- says there is still some black magic
involved in making them work. :-) )

The polyphase network was the subject of Michael Gingell's PhD
thesis in the UK. Material on that was seen on the Internet. Mike
is a USA resident now (or was a couple years ago when he had a
website...has a US ham callsign, too). A Japanese ham got busy
on that polyphase network and came up with an optimum set of
component values. That was published in QEX. Hans Summers'
website has links to all those.

Gabor Temes is a familiar name to textbook thumbers. :-) There
isn't a lot of black magic associated with the Gingell network, but it is
a thorough #$%^!!!! to try and analyze with its busy interconnections.
I stole a few minutes of CPU time on the RCA corporate computer,
using their LECAP (a frequency-domain version of IBM's ECAP...the
SPICE thing hadn't been developed yet) back in the 1970s. It worked
as advertised with only 0 and 180 degree audio input, producing nice
relative quadrature phases on all four outputs. Was surprised!

Some scrounged parts, not well measured as to values, and a quick
open-air toss-together showed excellent broadband relative quadrature
with less than 1 degree error in the 'voice' bandspace. Checked that
with a time-interval function on a homebuilt frequency counter.

Thanks for all the advice Len... I'd be offering to take you to dinner by
now if you were halfway local!

Thank you but I'll just wave as my wife and I roll through Oregon
along I-5 about once a year from southern California to Puget Sound
area of Washington. :-) Want to bypass the usual clogging just
before crossing the river into WA and vice-versa.

Len Anderson
retired (from regular hours) electronic engineer person

Avery Fineman wrote:
You CAN realize a lowpass
function in DSP but the TI chip inputs probably needs some sort
of hardware lowpass filtering...?

An anti-alias filter at the very least! The TI chip in question is actually
a microcontroller rather than a DSP; I chose that based on desiring a low
power design and only requiring one chip rather than two (although I'd grant
you that there are some, e.g., SO-8 package serial interface ADCs out there
that almost don't count as another chip...) and, uh, because I already have
experience with it from other projects. It can sample up to ~200ksps, but I
was shooting for a much lower rather (perhaps some 20-50 ksps) since there
isn't much number crunching 'oomph' available. (I.e., no multipy-accumulate
instruction. In fact, no hardware multiplier at all! :-) I'll build the
mind-numbingly fast DSP monster receiver that can pull a signal 10dB beneath
the noise floor and turn it into CD quality audio once I get the simple ones
working.)

Here's a good hint on melding hardware with software using DSP:

"Scientist's and Engineer's Guide to Digital Signal Processing,"
by Stephen W. Smith, PhD, California Technical Publishing.

How very interesting -- I actually have a copy of that with from Analog
Devices (under a slightly different name); I always figured it was some
third party and not the application engineers at Analog Devices that wrote
it. Now I just have to crack the thing open...

Good luck on that.

It should be up and running within a month here.

I've seen it claimed in text, but no details, that a quasi-lock could
be obtained via voice, working on the harmonics of speech tones.
I'm not going to buy that until I see a demo.

I figured that's what they might have had in mind. Seems like a lot of
effort to avoid sending a pilot tone (but then again, what else to use the
aforementioned DSP for... oh, wait... fancy digital modulation schemes...
ok...)

---Joel

Avery Fineman wrote:
You CAN realize a lowpass
function in DSP but the TI chip inputs probably needs some sort
of hardware lowpass filtering...?

An anti-alias filter at the very least! The TI chip in question is actually
a microcontroller rather than a DSP; I chose that based on desiring a low
power design and only requiring one chip rather than two (although I'd grant
you that there are some, e.g., SO-8 package serial interface ADCs out there
that almost don't count as another chip...) and, uh, because I already have
experience with it from other projects. It can sample up to ~200ksps, but I
was shooting for a much lower rather (perhaps some 20-50 ksps) since there
isn't much number crunching 'oomph' available. (I.e., no multipy-accumulate
instruction. In fact, no hardware multiplier at all! :-) I'll build the
mind-numbingly fast DSP monster receiver that can pull a signal 10dB beneath
the noise floor and turn it into CD quality audio once I get the simple ones
working.)

Here's a good hint on melding hardware with software using DSP:

"Scientist's and Engineer's Guide to Digital Signal Processing,"
by Stephen W. Smith, PhD, California Technical Publishing.

How very interesting -- I actually have a copy of that with from Analog
Devices (under a slightly different name); I always figured it was some
third party and not the application engineers at Analog Devices that wrote
it. Now I just have to crack the thing open...

Good luck on that.

It should be up and running within a month here.

I've seen it claimed in text, but no details, that a quasi-lock could
be obtained via voice, working on the harmonics of speech tones.
I'm not going to buy that until I see a demo.

I figured that's what they might have had in mind. Seems like a lot of
effort to avoid sending a pilot tone (but then again, what else to use the
aforementioned DSP for... oh, wait... fancy digital modulation schemes...
ok...)

---Joel

Also regarding microcontrollers vs. DSPs: TI has an application note where
they use so-called wave filters (which I know next to nothing about other
than the name and that they can be designed to only require
additions/subtractions and not multiplications in their implementations and
hence execute quickly) for detecting DTMF. I thought it was pretty
clever...

---Joel