Elecraft

KX2/3 DSP AM Demodulation

Classic

List

Threaded

13 messages Options

K9MA

KX2/3 DSP AM Demodulation

It's been a long time since I studied signal theory in college, but I'm
curious as to how AM is demodulated in a direct conversion radio like
the KX2/3. (SSB and CW demodulation I understand.) Does anyone know, or
have a reference?

73,

Scott K9MA

--
Scott K9MA

[hidden email]

______________________________________________________________
Elecraft mailing list
Home: http://mailman.qth.net/mailman/listinfo/elecraft
Help: http://mailman.qth.net/mmfaq.htm
Post: mailto:[hidden email]

This list hosted by: http://www.qsl.net
Please help support this email list: http://www.qsl.net/donate.html
Message delivered to [hidden email]

k6dgw

Re: KX2/3 DSP AM Demodulation

A superhet converts the signal to an intermediate frequency [455 KHz?].
Sometimes it does it more than once, 2 or more IF's, many today have a
very high frequency first IF [70+ Mhz].

A DC receiver converts the signal once to audio. Regardless of the
mechanics [which these days can get very complex], that's really all
there is.

73,

Fred ["Skip"] K6DGW
Sparks NV DM09dn
Washoe County

On 9/9/2017 7:19 PM, K9MA wrote:

> It's been a long time since I studied signal theory in college, but
> I'm curious as to how AM is demodulated in a direct conversion radio
> like the KX2/3. (SSB and CW demodulation I understand.) Does anyone
> know, or have a reference?
>
>
> 73,
>
> Scott K9MA
>

______________________________________________________________
Elecraft mailing list
Home: http://mailman.qth.net/mailman/listinfo/elecraft
Help: http://mailman.qth.net/mmfaq.htm
Post: mailto:[hidden email]

This list hosted by: http://www.qsl.net
Please help support this email list: http://www.qsl.net/donate.html
Message delivered to [hidden email]

K9MA

Re: KX2/3 DSP AM Demodulation

On 9/9/2017 21:35, Fred Jensen wrote:
> A DC receiver converts the signal once to audio. Regardless of the
> mechanics [which these days can get very complex], that's really all
> there is.

Ah, but the only way direct conversion can convert AM to audio, without
some kind of fancy DSP stuff, is to phase lock the conversion oscillator
to the carrier. My KX2 doesn't sound like it's doing that when I tune
across a signal; it sounds more like a conventional superhet AM
receiver. I'm trying to figure out how you take the I and Q mixer
baseband ("audio") outputs of an AM signal and convert them to audio,
and in a way that works when the carrier is not exactly zero beat.

73,

Scott K9MA

--
Scott K9MA

[hidden email]

______________________________________________________________
Elecraft mailing list
Home: http://mailman.qth.net/mailman/listinfo/elecraft
Help: http://mailman.qth.net/mmfaq.htm
Post: mailto:[hidden email]

This list hosted by: http://www.qsl.net
Please help support this email list: http://www.qsl.net/donate.html
Message delivered to [hidden email]

Don Wilhelm

Re: KX2/3 DSP AM Demodulation

In reply to this post by K9MA

Scott,

The KX2/3 is not a direct conversion receiver/transmitter.
Instead, it has a quadrature mixer which converts the incoming RF into
baseband (or 8kHz IF) I/Q streams which is then sent to analog to
digital converter. The output of the A to D converter is a 'bunch of
bits', which can be manipulated by mathematical processes (DSP) to
poduce most anything you wish.

So the KX2/3 receiver is similar to a SoftRock receiver which feeds its
output I/Q streams to a computer soundcard for DSP processing. With the
KX2/3, the soundcard and computer are contained inside the box (no
computer needed).

That is the best I can explain "how it works".
If your study of signal theory also included Digital Signal Processing,
then look to the DSP algorithms for modulation/demodulation of all forms
of RF signals. AM, CW, SSB, FM are all just mathematical formulas by
which to manipulate data.

The DSP process is more akin to the phasing method of detection. If you
want an all analog comparison, look at the phasing receivers that Rick
Campbell KK7B has designed. His receivers did "DSP" with analog
circuits, and the weakness of that approach is only that analog devices
have component tolerances, while mathematical formulas can produce exact
results without those tolerances.

73,
Don W3FPR

On 9/9/2017 10:19 PM, K9MA wrote:
> It's been a long time since I studied signal theory in college, but I'm
> curious as to how AM is demodulated in a direct conversion radio like
> the KX2/3. (SSB and CW demodulation I understand.) Does anyone know, or
> have a reference?
______________________________________________________________
Elecraft mailing list
Home: http://mailman.qth.net/mailman/listinfo/elecraft
Help: http://mailman.qth.net/mmfaq.htm
Post: mailto:[hidden email]

This list hosted by: http://www.qsl.net
Please help support this email list: http://www.qsl.net/donate.html
Message delivered to [hidden email]

Don Wilhelm

Re: KX2/3 DSP AM Demodulation

In reply to this post by K9MA

Scott,

The KX2/3 does not convert directly to audio.
It converts the RF to "baseband" which is a zero frequency IF (in the
KX3). The KX2 has the "8kHz shift" built in at all times, so the IF is
8kHz rather than baseband.
The KX3 can shift to the 8kHz IF if there is breakthrough of AM
broadcast stations.

Again, these are not Direct Conversion receivers.

73,
Don W3FPR

On 9/9/2017 10:46 PM, K9MA wrote:

> On 9/9/2017 21:35, Fred Jensen wrote:
>> A DC receiver converts the signal once to audio. Regardless of the
>> mechanics [which these days can get very complex], that's really all
>> there is.
>
> Ah, but the only way direct conversion can convert AM to audio, without
> some kind of fancy DSP stuff, is to phase lock the conversion oscillator
> to the carrier. My KX2 doesn't sound like it's doing that when I tune
> across a signal; it sounds more like a conventional superhet AM
> receiver. I'm trying to figure out how you take the I and Q mixer
> baseband ("audio") outputs of an AM signal and convert them to audio,
> and in a way that works when the carrier is not exactly zero beat.

K9MA

Re: KX2/3 DSP AM Demodulation

In reply to this post by K9MA

Never mind: I think I found the answer:

https://www.tjhsst.edu/~rlatimer/techlab/Gelbpaper04.pdf

(The explanation isn't that clear, but I worked through the trig a bit
more rigorously, and got the same answer.)

Basically, the modulating signal, m, is the square root of the sum of
the squares of I and Q.

So really, all the DSP has to do is a little simple arithmetic.

I should have been able to figure that out.

73,

Scott K9MA

On 9/9/2017 21:19, K9MA wrote:

--
Scott K9MA

[hidden email]

______________________________________________________________
Elecraft mailing list
Home: http://mailman.qth.net/mailman/listinfo/elecraft
Help: http://mailman.qth.net/mmfaq.htm
Post: mailto:[hidden email]

This list hosted by: http://www.qsl.net
Please help support this email list: http://www.qsl.net/donate.html
Message delivered to [hidden email]

Alan Bloom

Re: KX2/3 DSP AM Demodulation

Right, the amplitude of the signal is just the length of the I/Q vector,
which is sqrt(I^2 + Q^2). If the receiver is a little off-frequency,
the I/Q vector rotates (because the phase is constantly changing) but
the amplitude is still just the length of the vector.

Alan N1AL

On 09/09/2017 08:19 PM, K9MA wrote:

> Never mind: I think I found the answer:
>
> https://www.tjhsst.edu/~rlatimer/techlab/Gelbpaper04.pdf
>
> (The explanation isn't that clear, but I worked through the trig a bit
> more rigorously, and got the same answer.)
>
> Basically, the modulating signal, m, is the square root of the sum of
> the squares of I and Q.
>
> So really, all the DSP has to do is a little simple arithmetic.
>
> I should have been able to figure that out.
>
> 73,
>
> Scott K9MA
>
>
> On 9/9/2017 21:19, K9MA wrote:
>> It's been a long time since I studied signal theory in college, but
>> I'm curious as to how AM is demodulated in a direct conversion radio
>> like the KX2/3. (SSB and CW demodulation I understand.) Does anyone
>> know, or have a reference?
>>
>>
>> 73,
>>
>> Scott K9MA
>>
>

K9MA

Re: KX2/3 DSP AM Demodulation

In reply to this post by Don Wilhelm

On 9/9/2017 22:13, Don Wilhelm wrote:
> The KX2/3 does not convert directly to audio.
> It converts the RF to "baseband" which is a zero frequency IF (in the
> KX3). The KX2 has the "8kHz shift" built in at all times, so the IF
> is 8kHz rather than baseband.
> The KX3 can shift to the 8kHz IF if there is breakthrough of AM
> broadcast stations.
>
> Again, these are not Direct Conversion receivers.

I didn't realize the KX2 used an 8 kHz IF, though I see it is mentioned
in the manual. I guess that means the image is 16 kHz away from the
desired signal, and is suppressed by the I/Q demodulation, as in a
phasing transmitter. It does explain why the post mixer amplifier
cutoff frequencies are so high.

However, is not the KX3, when it's using the zero frequency IF really a
direct conversion receiver, just with quadrature mixers so one side of
zero beat can be suppressed? Perhaps the term "Direct Conversion" is
avoided because of the lack of opposite sideband rejection in
conventional DC receivers.

Just out of curiosity, why does the KX3 not use the 8 kHz IF all the
time, as does the KX2?

73,

Scott K9MA

--
Scott K9MA

[hidden email]

______________________________________________________________
Elecraft mailing list
Home: http://mailman.qth.net/mailman/listinfo/elecraft
Help: http://mailman.qth.net/mmfaq.htm
Post: mailto:[hidden email]

This list hosted by: http://www.qsl.net
Please help support this email list: http://www.qsl.net/donate.html
Message delivered to [hidden email]

vk2rq

Re: KX2/3 DSP AM Demodulation

It's so the roofing filter option can be used (using weaver demodulation).

73 de Matt VK2RQ

> On 10 Sep 2017, at 1:39 pm, K9MA <[hidden email]> wrote:
>
>> On 9/9/2017 22:13, Don Wilhelm wrote:
>> The KX2/3 does not convert directly to audio.
>> It converts the RF to "baseband" which is a zero frequency IF (in the KX3). The KX2 has the "8kHz shift" built in at all times, so the IF is 8kHz rather than baseband.
>> The KX3 can shift to the 8kHz IF if there is breakthrough of AM broadcast stations.
>>
>> Again, these are not Direct Conversion receivers.
>
> I didn't realize the KX2 used an 8 kHz IF, though I see it is mentioned in the manual. I guess that means the image is 16 kHz away from the desired signal, and is suppressed by the I/Q demodulation, as in a phasing transmitter. It does explain why the post mixer amplifier cutoff frequencies are so high.
>
> However, is not the KX3, when it's using the zero frequency IF really a direct conversion receiver, just with quadrature mixers so one side of zero beat can be suppressed? Perhaps the term "Direct Conversion" is avoided because of the lack of opposite sideband rejection in conventional DC receivers.
>
> Just out of curiosity, why does the KX3 not use the 8 kHz IF all the time, as does the KX2?
>
> 73,
>
> Scott K9MA
>
> --
> Scott K9MA
>
> [hidden email]
>
> ______________________________________________________________
> Elecraft mailing list
> Home: http://mailman.qth.net/mailman/listinfo/elecraft
> Help: http://mailman.qth.net/mmfaq.htm
> Post: mailto:[hidden email]
>
> This list hosted by: http://www.qsl.net
> Please help support this email list: http://www.qsl.net/donate.html
> Message delivered to [hidden email]

Don Wilhelm

Re: KX2/3 DSP AM Demodulation

In reply to this post by K9MA

Scott,

The advantage of using the baseband IF is that one can use a baseband
filter to provide a roofing filter. Such filters get more complex if
used at 8kHz.
It also allows RIT/XIT excursion that is equal on each side of the
carrier frequency. In the KX3, it is +/- 15kHz, while with the 8kHz
shift, the RIT/XIT range is +7 to -23kHz.

What is termed "Direct conversion" is a detector/mixer that goes from RF
directly to a single audio stream - it hears equally well on both the
desired and undesired side of the carrier.
In a Phasing or DSP based receiver, there are 2 outputs from the mixer -
out of phase by 90 degrees to each other. That allows the opposite
sideband to be suppressed. The process is done mathematically whether
by DSP techniques or analog components as in the KK7B "R2" or the very
old phasing detector transmitters and receivers available in the 1960s.
Central Electronics made some (including the "Signal Slicer" for a
receiver adapter and worked at the 455kHz IF common in receivers of that
day).

73,
Don W3FPR

On 9/9/2017 11:39 PM, K9MA wrote:

> I didn't realize the KX2 used an 8 kHz IF, though I see it is
> mentioned in the manual. I guess that means the image is 16 kHz away
> from the desired signal, and is suppressed by the I/Q demodulation, as
> in a phasing transmitter. It does explain why the post mixer
> amplifier cutoff frequencies are so high.
>
> However, is not the KX3, when it's using the zero frequency IF really
> a direct conversion receiver, just with quadrature mixers so one side
> of zero beat can be suppressed? Perhaps the term "Direct Conversion"
> is avoided because of the lack of opposite sideband rejection in
> conventional DC receivers.
>
> Just out of curiosity, why does the KX3 not use the 8 kHz IF all the
> time, as does the KX2?

k6dgw

Re: KX2/3 DSP AM Demodulation

In reply to this post by K9MA

Your KX2 isn't really a DC receiver, it has an 8KHz IF which is then
digitized and demodulated in a digital computer executing the demod
algorithms to produce baseband output.

But the original question was simply "what is direct conversion?" and
the answer is, you convert the incoming signal to baseband ... directly
... one time. What you do with the baseband after that is a separate
matter. And, you can hear SSB, CW, and AM just fine with a
free-running, reasonably stable LO.

My experience listening to an analog direct-conversion receiver in a low
ambient noise environment is that the RX noise is uncharacteristically
low, almost to the point that I thought something was wrong, like no
antenna. Then, tuning across a signal, it just sort of pops up at the
expected volume. Since I've only done this with one receiver, it's hard
to draw meaningful conclusions however. [:-)

73,

Fred ["Skip"] K6DGW
Sparks NV DM09dn
Washoe County

On 9/9/2017 7:46 PM, K9MA wrote:

> On 9/9/2017 21:35, Fred Jensen wrote:
>> A DC receiver converts the signal once to audio. Regardless of the
>> mechanics [which these days can get very complex], that's really all
>> there is.
>
> Ah, but the only way direct conversion can convert AM to audio,
> without some kind of fancy DSP stuff, is to phase lock the conversion
> oscillator to the carrier. My KX2 doesn't sound like it's doing that
> when I tune across a signal; it sounds more like a conventional
> superhet AM receiver. I'm trying to figure out how you take the I and
> Q mixer baseband ("audio") outputs of an AM signal and convert them to
> audio, and in a way that works when the carrier is not exactly zero beat.
>
> 73,
>
> Scott K9MA
>

K9MA

Re: KX2/3 DSP AM Demodulation

In reply to this post by Don Wilhelm

Don,

OK, I'll turn that around, and ask why the KX2 does NOT convert directly
to baseband, rather than to 8 kHz? You said something about AM signals
breaking in. What's that all about?

Call it what you will, but I still maintain that a receiver, like the
KX3, which converts directly to baseband is really a direct conversion
receiver, just using quadrature mixers so the opposite sideband can be
suppressed. I suppose you could call it a phasing direct conversion
receiver. I can see, however, why it might not be referred to as DC to
avoid the assumption that it receives both sidebands, but that's just
marketing.

73,

Scott K9MA

On 9/10/2017 06:54, Don Wilhelm wrote:

> Scott,
>
> The advantage of using the baseband IF is that one can use a baseband
> filter to provide a roofing filter. Such filters get more complex if
> used at 8kHz.
> It also allows RIT/XIT excursion that is equal on each side of the
> carrier frequency. In the KX3, it is +/- 15kHz, while with the 8kHz
> shift, the RIT/XIT range is +7 to -23kHz.
>
> What is termed "Direct conversion" is a detector/mixer that goes from
> RF directly to a single audio stream - it hears equally well on both
> the desired and undesired side of the carrier.
> In a Phasing or DSP based receiver, there are 2 outputs from the mixer
> - out of phase by 90 degrees to each other. That allows the opposite
> sideband to be suppressed. The process is done mathematically whether
> by DSP techniques or analog components as in the KK7B "R2" or the very
> old phasing detector transmitters and receivers available in the
> 1960s. Central Electronics made some (including the "Signal Slicer"
> for a receiver adapter and worked at the 455kHz IF common in receivers
> of that day).
>
> 73,
> Don W3FPR
>
> On 9/9/2017 11:39 PM, K9MA wrote:
>> I didn't realize the KX2 used an 8 kHz IF, though I see it is
>> mentioned in the manual. I guess that means the image is 16 kHz away
>> from the desired signal, and is suppressed by the I/Q demodulation,
>> as in a phasing transmitter. It does explain why the post mixer
>> amplifier cutoff frequencies are so high.
>>
>> However, is not the KX3, when it's using the zero frequency IF really
>> a direct conversion receiver, just with quadrature mixers so one side
>> of zero beat can be suppressed? Perhaps the term "Direct Conversion"
>> is avoided because of the lack of opposite sideband rejection in
>> conventional DC receivers.
>>
>> Just out of curiosity, why does the KX3 not use the 8 kHz IF all the
>> time, as does the KX2?
>

Don Wilhelm

Re: KX2/3 DSP AM Demodulation

Scott,

The fact that it produces I and Q outputs from the quadrature mixer puts
it in the class of SDR transceivers, which are a different breed than
Direct Conversion.
You are free to call it what you will, but the rest of the world refers
to that as an SDR.

The 8kHz shift allows the opposite sideband and other garbage to be 16
kHz away from the desired signal where it is easily dealt with - no
interfering results.

73,
Don W3FPR

On 9/10/2017 10:51 PM, K9MA wrote:

> Don,
>
> OK, I'll turn that around, and ask why the KX2 does NOT convert
> directly to baseband, rather than to 8 kHz? You said something about
> AM signals breaking in. What's that all about?
>
> Call it what you will, but I still maintain that a receiver, like the
> KX3, which converts directly to baseband is really a direct conversion
> receiver, just using quadrature mixers so the opposite sideband can be
> suppressed. I suppose you could call it a phasing direct conversion
> receiver. I can see, however, why it might not be referred to as DC
> to avoid the assumption that it receives both sidebands, but that's
> just marketing.
>
> 73,
>
> Scott K9MA
>
> On 9/10/2017 06:54, Don Wilhelm wrote:
>> Scott,
>>
>> The advantage of using the baseband IF is that one can use a baseband
>> filter to provide a roofing filter. Such filters get more complex if
>> used at 8kHz.
>> It also allows RIT/XIT excursion that is equal on each side of the
>> carrier frequency. In the KX3, it is +/- 15kHz, while with the 8kHz
>> shift, the RIT/XIT range is +7 to -23kHz.
>>
>> What is termed "Direct conversion" is a detector/mixer that goes from
>> RF directly to a single audio stream - it hears equally well on both
>> the desired and undesired side of the carrier.
>> In a Phasing or DSP based receiver, there are 2 outputs from the
>> mixer - out of phase by 90 degrees to each other. That allows the
>> opposite sideband to be suppressed. The process is done
>> mathematically whether by DSP techniques or analog components as in
>> the KK7B "R2" or the very old phasing detector transmitters and
>> receivers available in the 1960s. Central Electronics made some
>> (including the "Signal Slicer" for a receiver adapter and worked at
>> the 455kHz IF common in receivers of that day).
>>
>> 73,
>> Don W3FPR
>>
>> On 9/9/2017 11:39 PM, K9MA wrote:
>>> I didn't realize the KX2 used an 8 kHz IF, though I see it is
>>> mentioned in the manual. I guess that means the image is 16 kHz
>>> away from the desired signal, and is suppressed by the I/Q
>>> demodulation, as in a phasing transmitter. It does explain why the
>>> post mixer amplifier cutoff frequencies are so high.
>>>
>>> However, is not the KX3, when it's using the zero frequency IF
>>> really a direct conversion receiver, just with quadrature mixers so
>>> one side of zero beat can be suppressed? Perhaps the term "Direct
>>> Conversion" is avoided because of the lack of opposite sideband
>>> rejection in conventional DC receivers.
>>>
>>> Just out of curiosity, why does the KX3 not use the 8 kHz IF all the
>>> time, as does the KX2?
>>
>