From: "Joel Kolstad" on Mon, Mar 5
2007 9:24 am

"OH1GTF" wrote in message

One thing I want to add. I have built tayloe detector polyphase
network receiver before and I know,
that we need I and Q to cancel out the image. But how the heck AM and
FM is done?

I and Q give you phase information; sqrt(I^2+Q^2) gets you the amplitude of
the signal. Now... AM is not phase-modulated, right? So **assuming you can
synchronize your carrier to the incoming carrier**, "I" will be the original
signal and Q will be zero, so sqrt(I^2+Q^2)=I -- poof, only I needed! If you
chose Q, you'd just get zero (ideally) out of the receiver... but since you're
presumably in control of the local oscillator, you can just advance or retard
it 90 degrees and now Q is in-phase with the signal and I is zero. Hence AM
can be made to work with either I or Q... although it's not really
recommended, since -- if you have an IQ demodulator anyway -- you can build
SSB receivers as well, which is useful.

Good post, Joel. Let me emphasize some more simplistic points
for other readers:

BOTH I (in-phase) and Q (quadrature) carry the SAME amplitude
variation as modulation, ergo the AM information can be taken
from either one if desired.

Selection of sideband for AM uses the phase DIFFERENCE between
I and Q that, combined with a wideband audio relative phase-
shift network in a linear (algebraic addition/subtraction) mixing
circuit (just op-amps), selects the desired AM sideband.

The tricky part is that "synchronizing to the incoming carrier" bit: If the
receiver and transmitter have the exact same frequency but phase offsets of X
degrees, the result is that I receives the original signal mulitplied by
cos(X) and Q receives the original signal multiplied by sin(X). (This is just
the general case of what I described above where things were 90 degrees out of
phase.)

Heh, heh, while a concise statement of what it is, the above
is going to be a snow job to those not familiar enough with
more-advanced math. I know it got me a few decades ago and
I just finished explaining that to another last month (off-
line with the advantage of some scratch paper to show the
relationships).

Locking the AM carrier to the local detector sub-circuit
oscillator can be done by simply running the whole IF
signal into a LIMITER to blot out the sidebands. The resulting
limited carrier is then used as a reference to lock the local
(detector) oscillator. "Lock" may be simplistic since the
old Motorola MC1330 video detector simply limits the carrier
internally, uses a single L-C external circuit to keep the
internal I and Q references at quadrature, and the two
internal "product detectors" output is linearly added
internally. Blessedly simple circuit, just one 8-pin DIP
and a very ordinary L-C that isn't critical in tuning. I used
that in a bank of 8 pulse detectors having carriers up to
62 MHz with no problems (even crammed into a thin space).

FM detection by similar, but NOT identical means, to AM.
The modulation equations for FM and PM don't allow that
similarity since sideband content is not really close to
AM. That is done in the more familiar "ratio detector"
or "quadrature detector" (sometimes an alternate name)
found in consumer electronics SOCs for FM receivers.
Try as I might this morning I couldn't put together a
text-only simple description of how that critter works. :-(

"Synchronous detection" of AM can be done on the same
model of the internals of the Motorola MC1330 IC with,
perhaps some more finesse applied to the detector's
local oscillator to remove some of the noise on low-level
input signals; not a terribly-important thing nor precise
since the oscillator's lock DC signal can be lowpassed
to around a half-second time constant in the loop filter.

Notice that phase and amplituide, while connected by phase=arctan(Q/I) and
amplitude=sqrt(I^2+Q^2), are two separate, uniquely "identificable" "things"
that you can transmit. This is taken advantage of in, e.g., "compatble" AM
stereo broadcast standards: In Motorola C-QUAM, for instance, I is set to
1+L+R whereas Q is set to L-R. If you run through the math, the amplitude of
this is not 1+L+R, but it is "close enough" if L-R is relatively small, and
hence compatibility with traditional (envelope) receivers is maintained, while
allowing a synchronous receiver to dig out the full stereo information.

C-QUAM follows the FM Stereo system where the demodulated
audio is the sum of Left plus Right. One hears (directly)
both sources for monophonic audio. The supersonic audio
carries Left minus Right audio. Linearly added and
subtracted, the Left plus Left and Right plus Right gets
the individual "ears" separated. The C-QUAM audio does
have a slight stereo distortion on odd selective conditions
that favor one sideband over another locally, typically
the effect of electric power distribution lines close to
the receiver site. Typical on my particular street. :-(

73,

Thanks for the additions, Len!

wrote in message
ps.com...
Heh, heh, while a concise statement of what it is, the above
is going to be a snow job to those not familiar enough with
more-advanced math.

Many textbooks and professors tell you the "bad news" about a phase difference
and that unwanted cos(theta) term... but they seldom mention that what "I
taketh," "Q giveth," and that you haven't actually lost anything yet -- if you
have both I and Q available -- just because you aren't phase-locked. I
suppose this is because, at that point, most textbooks are still hundreds of
pages away from discussing I-Q methods.

A lot of the texts make quite a mire out of the whole "arbitrary band-limited
signal impressed upon a carrier" bit. I think most people are almost better
off ignoring it initially (it seems to come up within the first few chapters
in many texts!) and then coming back to it once they have a good "feel" for
how phase modulation techniques work.

---Joel

Huge thanks Glenn, Joel, Paul and Len.

I had to read your mails 'few' times before I fully understood the
content. This subject acutally isn't anything you read at coffee break
at work within 5 minutes, hi!

I'm gonna stick of course with the IQ instead of using only other
component since I've build and IQ transceiver I was just pretty
amazed that I could have clear FM or AM audio out just by using one
component, but immediately notice the 6db change. But now, it's clear
to me, there's nothing special in that. The same amplitude can be
found in I and Q and OF COURSE we can demodulate AM or FM without the
second component.

Actually, Joel, your answers were pretty good for me, even with the
"more advanced" math, like Len said, hi. At first glance, I couldn't
get anything in to my head, but reading it few times more started to
reveal the math behind the whole thing. I'm really looking forward to
get more into the math and understanding the IQ demodulation and
modulation.

I had very good succestions even for the hardware stuff, but I think
I'll stick with the IQ rx/tx I just made. Actually, those HW ideas
also gave new aspects of thinking the way how IQ signals can be
demodulated in software. I also liked the C-QUAM information. Really
nice, because I didn't knew about the whole C-QUAM -thing.

Wildest dream would be of course, FULLY understand the demodulation
of any common modes and write my own software, but I've just done some
non-FFT releated stuff with Visual Basic, so not so promising

I want to thank everyone for the comments and advices, I had more
than I've read so far from the articles laying around the internet.
Always when ppl are talking to ppl, they communicate, but reading from
the books, that's just trying to figure out what a heck has been in
the mind of the writer. Ok, depends on the article, some might write
in a more common sense.

Just on question mo If I really want to dig into the math thing of
IQ modulation/demodulation, is there any articles you would like to
succest? I mean, I've never been that good in high tech math, so at
least the beginning of the article would have to be understood with
'basic' math.

73 de Janne, OH1GTF

PS. Yesterday my friend OH1CO brought RFspace's SDR-IQ transceiver to
the club (club evening) which he just bought. At our city club, we
have only 80m dipole hanging on the trees, but that didn't bother us,
we could copy wery interesting station on all bands at HF, all over
the world. I'm a 'bit' excited! I also received PSK station all over
the world @ 80 meters, with just few meters of wire in my shack
(powersdr and digipan).

On 8 Mar 2007 21:50:51 -0800, OH1GTF wrote in .com:

73 de Janne, OH1GTF

PS. Yesterday my friend OH1CO brought RFspace's SDR-IQ transceiver to
the club (club evening) which he just bought. At our city club, we
have only 80m dipole hanging on the trees, but that didn't bother us,
we could copy wery interesting station on all bands at HF, all over
the world. I'm a 'bit' excited! I also received PSK station all over
the world @ 80 meters, with just few meters of wire in my shack
(powersdr and digipan).

Janne,

I'm about to order an SDR-IQ board. I thought it was just a receiver.
If it's a transceiver, that's a big bonus.

--
Mike Andrews, W5EGO

Tired old sysadmin

Oh sorry Mike, my fault, I talked so much about my transceiver that
the word some how caught to my head. Yep, the SDR-IQ is rx only. Sorry
for the mix up!

73 de Janne, OH1GTF