I've been toying around with the idea of digitizing the entire AM commercial
broadcast band and -- mainly for my own edification -- doing demodulation in
a microcontroller or DSP. Although on one hand I'd claim I can then do cool
things like AM stereo demodulation, on the other hand I do recognize such
coolness is lost in that digitizing the _entire AM band at once_ from
550-1600KHz only allows me to adjust the gain of an RF amplifier to the
point where _the sum of the signals from all the stations_ doesn't overload
an ADC. In other words, this approach will only work to pick out relatively
strong stations if I still want anything resembling a decent SNR, correct?

To get around this problem, then, I was thinking of using an NE602 mixer to
select just the station I wanted and then following its output with an
adjustable RF amp. If I go for the direct conversion approach, then, I can
follow it with a variable gain baseband amplifier and still hear the weak
stations with relatively good fideilty.

Is this a reasonable approach? Is it true that low (...to moderate?)
quality receivers can get away with just feeding an antenna's output to a
fixed gain broadband amplifier (or even just a matching/bandpass filtering
network) followed by a mixer and _then_ an adjustable gain element? (As
opposed to high quality receivers that appear to have tuned RF amplifier
front ends?)

The idea of digitizing an entire band at once seems to look far less
promising if your goal is to dig out weak stations...

Thanks,
---Joel Kolstad

(I also find it quite amusing to think I can build an evelope detector in
software to test things out prior to building fancy synchronous detection
routines!)

On Sat, 18 Oct 2003 14:06:36 -0700, "Joel Kolstad"
wrote:

I've been toying around with the idea of digitizing the entire AM commercial
broadcast band and -- mainly for my own edification -- doing demodulation in
a microcontroller or DSP. Although on one hand I'd claim I can then do cool
things like AM stereo demodulation, on the other hand I do recognize such
coolness is lost in that digitizing the _entire AM band at once_ from
550-1600KHz only allows me to adjust the gain of an RF amplifier to the
point where _the sum of the signals from all the stations_ doesn't overload
an ADC. In other words, this approach will only work to pick out relatively
strong stations if I still want anything resembling a decent SNR, correct?

To get around this problem, then, I was thinking of using an NE602 mixer to
select just the station I wanted and then following its output with an
adjustable RF amp. If I go for the direct conversion approach, then, I can
follow it with a variable gain baseband amplifier and still hear the weak
stations with relatively good fideilty.

Is this a reasonable approach? Is it true that low (...to moderate?)
quality receivers can get away with just feeding an antenna's output to a
fixed gain broadband amplifier (or even just a matching/bandpass filtering
network) followed by a mixer and _then_ an adjustable gain element? (As
opposed to high quality receivers that appear to have tuned RF amplifier
front ends?)

The idea of digitizing an entire band at once seems to look far less
promising if your goal is to dig out weak stations...

Thanks,
---Joel Kolstad

(I also find it quite amusing to think I can build an evelope detector in
software to test things out prior to building fancy synchronous detection
routines!)


How about some front-end selectivity using a varactor tuned circuit?
That allows for tuning under software control using a D-A converter.

A GE SuperRadio 3 could be used as a pre-built prototype. It uses 3
varactors (one for tuned RF stage) instead of 3 mechanical variable
capacitors. You could take either the tuned RF stage output or IF
output.

Have fun,
Jim

On Sat, 18 Oct 2003 14:06:36 -0700, "Joel Kolstad"
wrote:

I've been toying around with the idea of digitizing the entire AM commercial
broadcast band and -- mainly for my own edification -- doing demodulation in
a microcontroller or DSP. Although on one hand I'd claim I can then do cool
things like AM stereo demodulation, on the other hand I do recognize such
coolness is lost in that digitizing the _entire AM band at once_ from
550-1600KHz only allows me to adjust the gain of an RF amplifier to the
point where _the sum of the signals from all the stations_ doesn't overload
an ADC. In other words, this approach will only work to pick out relatively
strong stations if I still want anything resembling a decent SNR, correct?

To get around this problem, then, I was thinking of using an NE602 mixer to
select just the station I wanted and then following its output with an
adjustable RF amp. If I go for the direct conversion approach, then, I can
follow it with a variable gain baseband amplifier and still hear the weak
stations with relatively good fideilty.

Is this a reasonable approach? Is it true that low (...to moderate?)
quality receivers can get away with just feeding an antenna's output to a
fixed gain broadband amplifier (or even just a matching/bandpass filtering
network) followed by a mixer and _then_ an adjustable gain element? (As
opposed to high quality receivers that appear to have tuned RF amplifier
front ends?)

The idea of digitizing an entire band at once seems to look far less
promising if your goal is to dig out weak stations...

Thanks,
---Joel Kolstad

(I also find it quite amusing to think I can build an evelope detector in
software to test things out prior to building fancy synchronous detection
routines!)


How about some front-end selectivity using a varactor tuned circuit?
That allows for tuning under software control using a D-A converter.

A GE SuperRadio 3 could be used as a pre-built prototype. It uses 3
varactors (one for tuned RF stage) instead of 3 mechanical variable
capacitors. You could take either the tuned RF stage output or IF
output.

Have fun,
Jim

On Sat, 18 Oct 2003 14:06:36 -0700, "Joel Kolstad"
wrote:

I've been toying around with the idea of digitizing the entire AM
commercial
broadcast band and -- mainly for my own edification -- doing demodulation
in
a microcontroller or DSP. Although on one hand I'd claim I can then do
cool
things like AM stereo demodulation, on the other hand I do recognize such
coolness is lost in that digitizing the _entire AM band at once_ from
550-1600KHz only allows me to adjust the gain of an RF amplifier to the
point where _the sum of the signals from all the stations_ doesn't
overload
an ADC. In other words, this approach will only work to pick out
relatively
strong stations if I still want anything resembling a decent SNR,
correct?

To get around this problem, then, I was thinking of using an NE602 mixer
to
select just the station I wanted and then following its output with an
adjustable RF amp. If I go for the direct conversion approach, then, I
can
follow it with a variable gain baseband amplifier and still hear the weak
stations with relatively good fideilty.

Is this a reasonable approach? Is it true that low (...to moderate?)
quality receivers can get away with just feeding an antenna's output to a
fixed gain broadband amplifier (or even just a matching/bandpass
filtering
network) followed by a mixer and _then_ an adjustable gain element? (As
opposed to high quality receivers that appear to have tuned RF amplifier
front ends?)

The idea of digitizing an entire band at once seems to look far less
promising if your goal is to dig out weak stations...

Shouldn't you look at the TenTec RX320? It superhets to 12kHz and then does
the final selectivity and detection in DSP. TenTec will modify old RX320s to
bring out the 12 kHz IF to allow playing using a sound card (say DRM).

Or maybe the Software Defined Radio articles by Youngblood in QEX almost a
year ago. He is selling card sets. A lot of his complexity is in a bank of
switchable filters to cover the "zero" to 50 MHz range.

Or the Motorola "Symphony" chipset for AM/FM broadcast.

Or similar software radio chips from Visteon/Ford.

Or a paper in the Proceedings of the 1999 Central States VHF Conference* by
Bob Stenowski, Director, Advanced Comm Engineering, Rockwell Collins. He
pointed out that you would need an AtoD converter with more than 25 bits
resolution/linearity to build a "DC to Daylight" receiver without frontend
tuning / filters. He shows block diagrams of the Collins 95S-1A which covers
5kHz to 2Ghz, but has 9 bandpass filters to cover the range. The 5kHz to
30MHz range is upconverted with a variable LO to 51.2MHz and then handled
like the VHF signals.

* available from ARRL


You have an interesting project.

On Sat, 18 Oct 2003 14:06:36 -0700, "Joel Kolstad"
wrote:

I've been toying around with the idea of digitizing the entire AM
commercial
broadcast band and -- mainly for my own edification -- doing demodulation
in
a microcontroller or DSP. Although on one hand I'd claim I can then do
cool
things like AM stereo demodulation, on the other hand I do recognize such
coolness is lost in that digitizing the _entire AM band at once_ from
550-1600KHz only allows me to adjust the gain of an RF amplifier to the
point where _the sum of the signals from all the stations_ doesn't
overload
an ADC. In other words, this approach will only work to pick out
relatively
strong stations if I still want anything resembling a decent SNR,
correct?

To get around this problem, then, I was thinking of using an NE602 mixer
to
select just the station I wanted and then following its output with an
adjustable RF amp. If I go for the direct conversion approach, then, I
can
follow it with a variable gain baseband amplifier and still hear the weak
stations with relatively good fideilty.

Is this a reasonable approach? Is it true that low (...to moderate?)
quality receivers can get away with just feeding an antenna's output to a
fixed gain broadband amplifier (or even just a matching/bandpass
filtering
network) followed by a mixer and _then_ an adjustable gain element? (As
opposed to high quality receivers that appear to have tuned RF amplifier
front ends?)

The idea of digitizing an entire band at once seems to look far less
promising if your goal is to dig out weak stations...

Shouldn't you look at the TenTec RX320? It superhets to 12kHz and then does
the final selectivity and detection in DSP. TenTec will modify old RX320s to
bring out the 12 kHz IF to allow playing using a sound card (say DRM).

Or maybe the Software Defined Radio articles by Youngblood in QEX almost a
year ago. He is selling card sets. A lot of his complexity is in a bank of
switchable filters to cover the "zero" to 50 MHz range.

Or the Motorola "Symphony" chipset for AM/FM broadcast.

Or similar software radio chips from Visteon/Ford.

Or a paper in the Proceedings of the 1999 Central States VHF Conference* by
Bob Stenowski, Director, Advanced Comm Engineering, Rockwell Collins. He
pointed out that you would need an AtoD converter with more than 25 bits
resolution/linearity to build a "DC to Daylight" receiver without frontend
tuning / filters. He shows block diagrams of the Collins 95S-1A which covers
5kHz to 2Ghz, but has 9 bandpass filters to cover the range. The 5kHz to
30MHz range is upconverted with a variable LO to 51.2MHz and then handled
like the VHF signals.

* available from ARRL


You have an interesting project.

Thanks for the responses so far...

The idea of digitizing the entire AM band would be a lot more appealing if
someone made one of those small serial interface ADCs that had a sample rate
in the 3-4MSps range. The best I've found so far is a 1.75MSps from Analog
Devices (AD7470, a 10 bit ADC), which -- even restricting the AM band to the
'traditional' 550-1550kHz instead of the slightly expanded 500?-1710kHz --
obviously isn't adequate to sample the entire 1MHz spectrum.

Hence I think I'm going to try an approach along the lines of a
micrcontroller's PWM output feeding a Linear Tech. LTC1799 square wave
oscillator, filtering its harmonics a bit, feeding it to an NE602A mixer
followed by a (means yet to be determined) adjustable gain IF amp, and then
the ADC on the microcontroller. For the audio output I the TI TLV320DAC23
is a high quality stereo DAC with built in 100mW audio amplifiers and serial
microcontroller connectivity.

Of course this is all going to have to fit into an Altoid's tin, right? :-)

---Joel Kolstad

Thanks for the responses so far...

The idea of digitizing the entire AM band would be a lot more appealing if
someone made one of those small serial interface ADCs that had a sample rate
in the 3-4MSps range. The best I've found so far is a 1.75MSps from Analog
Devices (AD7470, a 10 bit ADC), which -- even restricting the AM band to the
'traditional' 550-1550kHz instead of the slightly expanded 500?-1710kHz --
obviously isn't adequate to sample the entire 1MHz spectrum.

Hence I think I'm going to try an approach along the lines of a
micrcontroller's PWM output feeding a Linear Tech. LTC1799 square wave
oscillator, filtering its harmonics a bit, feeding it to an NE602A mixer
followed by a (means yet to be determined) adjustable gain IF amp, and then
the ADC on the microcontroller. For the audio output I the TI TLV320DAC23
is a high quality stereo DAC with built in 100mW audio amplifiers and serial
microcontroller connectivity.

Of course this is all going to have to fit into an Altoid's tin, right? :-)

---Joel Kolstad