Direct conversion receivers for AM
 

I had this crazy idea to attempt to build a direct conversion receiver for
regular old AM (i.e., two sidebands, carrier included). It eventually
occurred to me, however, that there's the very significant problem of
synchronizing the LO to the incoming RF carrier. In diagrams I've seen,
normally a PLL is used at the IF frequency (455kHz being common for AM
receivers, of course) to lock the two together. Is there anything
comparable one can do with direct conversion? It seems that if your LO is
sync'd with the incoming carrier, you'll have a very large DC component (the
AM carrier!)... if not, however, you still might get some DC that's been
aliased and while this will (should) always be less than the DC component
when you're locked, it gives no indication of which way your LO needs to
move to achieve lock.

So apparently what I'm really asking is... can one build a phase detector
that works at DC? Or does an AM direct conversion receiver necessarily
require an IF strip?

Thinking out loud,
---Joel Kolstad

I guess this is further to the followup I just posted in another
thread!

Consider the quadrature mixers. If your LO is in phase with the
incoming carrier on one, as you say, there will be a high DC coming
out of that mixer. But out of the other mixer at that time, you'll
see zero DC. If you move the phase just a bit one way, the quadrature
detector output will go up; move it the other way and the quadrature
detector output will go down. So, if you appropriately filter that DC
signal, you in theory will be able to lock the phase of the LO. The
problem? You'll need to be darned sure your mixer is really balanced!
If you are receiving, let's say, a 10 microvolt carrier, then you
only get around +/-10uV of DC to work with, and if the mixer is
imbalanced enough with no signal at all into the RF input to give you
that much DC, you'll be in trouble trying to "see" the DC from the
small carrier.

But not all is lost. As noted in the other thread, you may be able to
combine the outputs of the two quadrature detectors and use that.
What you won't get is rejection of noise that's in quadrature with the
transmitted carrier, which is an advantage of synchronous detection
(properly phase locked).

I've thought that if you have two broadcast stations whose carriers
are phase locked, it should be possible to find an antenna location
for receiving that puts the two carriers 90 degrees out of phase, at
the receiver, and you could listen to the two stations independently.
I have a spectrum analyzer that can resolve (very) small fractions of
a Hz at RF, and it's interesting to put a few feet of wire on it as an
antenna and set it up on a standard MW broadcast frequency with a 10Hz
span. Even in the presence of a strong local carrier, you can see
typically several other carriers in there, at least during nighttime
propagation.

Cheers,
Tom

"Joel Kolstad" wrote in message ...
I had this crazy idea to attempt to build a direct conversion receiver for
regular old AM (i.e., two sidebands, carrier included). It eventually
occurred to me, however, that there's the very significant problem of
synchronizing the LO to the incoming RF carrier. In diagrams I've seen,
normally a PLL is used at the IF frequency (455kHz being common for AM
receivers, of course) to lock the two together. Is there anything
comparable one can do with direct conversion? It seems that if your LO is
sync'd with the incoming carrier, you'll have a very large DC component (the
AM carrier!)... if not, however, you still might get some DC that's been
aliased and while this will (should) always be less than the DC component
when you're locked, it gives no indication of which way your LO needs to
move to achieve lock.

So apparently what I'm really asking is... can one build a phase detector
that works at DC? Or does an AM direct conversion receiver necessarily
require an IF strip?

Thinking out loud,
---Joel Kolstad

I guess this is further to the followup I just posted in another
thread!

Consider the quadrature mixers. If your LO is in phase with the
incoming carrier on one, as you say, there will be a high DC coming
out of that mixer. But out of the other mixer at that time, you'll
see zero DC. If you move the phase just a bit one way, the quadrature
detector output will go up; move it the other way and the quadrature
detector output will go down. So, if you appropriately filter that DC
signal, you in theory will be able to lock the phase of the LO. The
problem? You'll need to be darned sure your mixer is really balanced!
If you are receiving, let's say, a 10 microvolt carrier, then you
only get around +/-10uV of DC to work with, and if the mixer is
imbalanced enough with no signal at all into the RF input to give you
that much DC, you'll be in trouble trying to "see" the DC from the
small carrier.

But not all is lost. As noted in the other thread, you may be able to
combine the outputs of the two quadrature detectors and use that.
What you won't get is rejection of noise that's in quadrature with the
transmitted carrier, which is an advantage of synchronous detection
(properly phase locked).

I've thought that if you have two broadcast stations whose carriers
are phase locked, it should be possible to find an antenna location
for receiving that puts the two carriers 90 degrees out of phase, at
the receiver, and you could listen to the two stations independently.
I have a spectrum analyzer that can resolve (very) small fractions of
a Hz at RF, and it's interesting to put a few feet of wire on it as an
antenna and set it up on a standard MW broadcast frequency with a 10Hz
span. Even in the presence of a strong local carrier, you can see
typically several other carriers in there, at least during nighttime
propagation.

Cheers,
Tom

"Joel Kolstad" wrote in message ...
I had this crazy idea to attempt to build a direct conversion receiver for
regular old AM (i.e., two sidebands, carrier included). It eventually
occurred to me, however, that there's the very significant problem of
synchronizing the LO to the incoming RF carrier. In diagrams I've seen,
normally a PLL is used at the IF frequency (455kHz being common for AM
receivers, of course) to lock the two together. Is there anything
comparable one can do with direct conversion? It seems that if your LO is
sync'd with the incoming carrier, you'll have a very large DC component (the
AM carrier!)... if not, however, you still might get some DC that's been
aliased and while this will (should) always be less than the DC component
when you're locked, it gives no indication of which way your LO needs to
move to achieve lock.

So apparently what I'm really asking is... can one build a phase detector
that works at DC? Or does an AM direct conversion receiver necessarily
require an IF strip?

Thinking out loud,
---Joel Kolstad

Tom Bruhns wrote:
I guess this is further to the followup I just posted in another
thread!

Yep!

Consider the quadrature mixers. If your LO is in phase with the
incoming carrier on one, as you say, there will be a high DC coming
out of that mixer. But out of the other mixer at that time, you'll
see zero DC. If you move the phase just a bit one way, the quadrature
detector output will go up; move it the other way and the quadrature
detector output will go down. So, if you appropriately filter that DC
signal, you in theory will be able to lock the phase of the LO.

This makes sense to me if you stick, e.g., (1+s(t)) -- where s(t) has not DC
component -- on your I modulator and, say, t(t) -- where t(t) has no DC
component -- on your Q modulator. DC filtering the I output, then, should
directly provide cos(phi) -- where phi is the phase offset phase the RF and
LO carriers -- and DC filtering the Q output provides sin(phi) (which is
probably the more useful signal for the sake of feedback). (For regular AM
transmissions, t(t)=0... I'm just thinking ahead to AM stereo here.)

The
problem? You'll need to be darned sure your mixer is really balanced!

A pair of NE602s sitting side by side doesn't really cut it? :-) I was
actually thinking that the solution here is that you let the AGC run off of
the quadrature detector output (sqrt(s(t)^2+t(t)^2), assuming you removed DC
from both channels) and only when the signal strength is above a certain
level do you allow the phase locking circuitry to kick in and attempt to
null out the phase differences.

I've thought that if you have two broadcast stations whose carriers
are phase locked, it should be possible to find an antenna location
for receiving that puts the two carriers 90 degrees out of phase, at
the receiver, and you could listen to the two stations independently.

Good point!

Side note: In looking up quadrature detectors, I was amused to find that
someone has a patent on using one for standard commercial AM band reception.
Amazing...

For more on LO vs. RF carrier phasing differences, I'm going to follow up to
the other thread.

---Joel

Tom Bruhns wrote:
I guess this is further to the followup I just posted in another
thread!

Yep!

Consider the quadrature mixers. If your LO is in phase with the
incoming carrier on one, as you say, there will be a high DC coming
out of that mixer. But out of the other mixer at that time, you'll
see zero DC. If you move the phase just a bit one way, the quadrature
detector output will go up; move it the other way and the quadrature
detector output will go down. So, if you appropriately filter that DC
signal, you in theory will be able to lock the phase of the LO.

This makes sense to me if you stick, e.g., (1+s(t)) -- where s(t) has not DC
component -- on your I modulator and, say, t(t) -- where t(t) has no DC
component -- on your Q modulator. DC filtering the I output, then, should
directly provide cos(phi) -- where phi is the phase offset phase the RF and
LO carriers -- and DC filtering the Q output provides sin(phi) (which is
probably the more useful signal for the sake of feedback). (For regular AM
transmissions, t(t)=0... I'm just thinking ahead to AM stereo here.)

The
problem? You'll need to be darned sure your mixer is really balanced!

A pair of NE602s sitting side by side doesn't really cut it? :-) I was
actually thinking that the solution here is that you let the AGC run off of
the quadrature detector output (sqrt(s(t)^2+t(t)^2), assuming you removed DC
from both channels) and only when the signal strength is above a certain
level do you allow the phase locking circuitry to kick in and attempt to
null out the phase differences.

I've thought that if you have two broadcast stations whose carriers
are phase locked, it should be possible to find an antenna location
for receiving that puts the two carriers 90 degrees out of phase, at
the receiver, and you could listen to the two stations independently.

Good point!

Side note: In looking up quadrature detectors, I was amused to find that
someone has a patent on using one for standard commercial AM band reception.
Amazing...

For more on LO vs. RF carrier phasing differences, I'm going to follow up to
the other thread.

---Joel

"Joel Kolstad" wrote in message ...
I had this crazy idea to attempt to build a direct conversion receiver for
regular old AM (i.e., two sidebands, carrier included). It eventually
occurred to me, however, that there's the very significant problem of
synchronizing the LO to the incoming RF carrier. In diagrams I've seen,
normally a PLL is used at the IF frequency (455kHz being common for AM
receivers, of course) to lock the two together. Is there anything
comparable one can do with direct conversion? It seems that if your LO is
sync'd with the incoming carrier, you'll have a very large DC component (the
AM carrier!)... if not, however, you still might get some DC that's been
aliased and while this will (should) always be less than the DC component
when you're locked, it gives no indication of which way your LO needs to
move to achieve lock.

So apparently what I'm really asking is... can one build a phase detector
that works at DC? Or does an AM direct conversion receiver necessarily
require an IF strip?

Of course: this is how synchronous AM detection works.
They generate a clean, in-phase carrier and use that to
demodulate the incoming signal.

To make a direct conversion receiver on similar principles
would require a frequency-agile PLL. You could use a Costas
loop, for example. If I was to do it today I'd convert
to I/Q baseband and do the rest in DSP.

Laura Halliday VE7LDH "Que les nuages soient notre
Grid: CN89mg pied a terre..."
ICBM: 49 16.05 N 122 56.92 W - Hospital/Shafte

"Joel Kolstad" wrote in message ...
I had this crazy idea to attempt to build a direct conversion receiver for
regular old AM (i.e., two sidebands, carrier included). It eventually
occurred to me, however, that there's the very significant problem of
synchronizing the LO to the incoming RF carrier. In diagrams I've seen,
normally a PLL is used at the IF frequency (455kHz being common for AM
receivers, of course) to lock the two together. Is there anything
comparable one can do with direct conversion? It seems that if your LO is
sync'd with the incoming carrier, you'll have a very large DC component (the
AM carrier!)... if not, however, you still might get some DC that's been
aliased and while this will (should) always be less than the DC component
when you're locked, it gives no indication of which way your LO needs to
move to achieve lock.

So apparently what I'm really asking is... can one build a phase detector
that works at DC? Or does an AM direct conversion receiver necessarily
require an IF strip?

Of course: this is how synchronous AM detection works.
They generate a clean, in-phase carrier and use that to
demodulate the incoming signal.

To make a direct conversion receiver on similar principles
would require a frequency-agile PLL. You could use a Costas
loop, for example. If I was to do it today I'd convert
to I/Q baseband and do the rest in DSP.

Laura Halliday VE7LDH "Que les nuages soient notre
Grid: CN89mg pied a terre..."
ICBM: 49 16.05 N 122 56.92 W - Hospital/Shafte

The trick of phase locking to the IF definitely works. I believe it's
called synchronous detection. One trick that might allow for simpler
hardware would be to use a 5 kHz IF. (You'd need to use filtering or
phase cancellation to get rid of the unwanted product on the other
side of the desired signal.)

It should even be possible to use quadrature detection with the LO off
by 5 kHz, then feed both the I and Q channels into a DSP and "phase
lock" to the IF frequency in software.

A coworker of mine told me about still another DSP trick, where you
tune as close as you can to the carrier, but of course you're a few
hertz off. Looking at an I/Q diagram, the slightly off-frequency AM
signal will look like a rotating vector. So long as there isn't a lot
of noise, the DSP can keep track of the polarity as the phase rotates,
and recover the AM as the magnitude of the vector.

David
--
In electronics, no one thinks you're strange if you blow your PAL.

The trick of phase locking to the IF definitely works. I believe it's
called synchronous detection. One trick that might allow for simpler
hardware would be to use a 5 kHz IF. (You'd need to use filtering or
phase cancellation to get rid of the unwanted product on the other
side of the desired signal.)

It should even be possible to use quadrature detection with the LO off
by 5 kHz, then feed both the I and Q channels into a DSP and "phase
lock" to the IF frequency in software.

A coworker of mine told me about still another DSP trick, where you
tune as close as you can to the carrier, but of course you're a few
hertz off. Looking at an I/Q diagram, the slightly off-frequency AM
signal will look like a rotating vector. So long as there isn't a lot
of noise, the DSP can keep track of the polarity as the phase rotates,
and recover the AM as the magnitude of the vector.

David
--
In electronics, no one thinks you're strange if you blow your PAL.

Of course: this is how synchronous AM detection works.
They generate a clean, in-phase carrier and use that to
demodulate the incoming signal.

Laura & Joel-

I was just thinking that. I recall there used to be (may still be) a shortwave
broadcast station just above the 15 Meter band, that broadcasted with a partly
suppressed carrier. Apparently the SWL community has access to equipment with
Synchronous detection capability.

Listening to it on an AM receiver, it could be understood but was distorted.
It was easy to tune it on a Ham transceiver as if it was SSB, but tuning was
critical for music to sound good.

Addressing the original question about direct conversion, what if you used a
temperature compensated oscillator with a phase-locked-loop? If the assumption
could be made that the broadcast station will not drift, then you could tune
each station dead on with an RIT-like adjustment of the oscillator.

73, Fred, K4DII