The 'phasing method' of single sideband demodulation

Hi guys,
  Some weeks ago I posted a message asking for help 
in understanding the 'phasing method' of single 
sideband (SSB) demodulation.

You guys helped me out, and I learned so much that 
I wrote a blog on that SSB demodulation method.  
In case anyone's interested, the blog is at:

http://www.dsprelated.com/showarticle/176.php

See Ya',
[-Rick-]

On Wed, 08 Aug 2012 15:04:38 -0700, Rick Lyons
<R.Lyons@_BOGUS_ieee.org> wrote:

>
>Hi guys,
>  Some weeks ago I posted a message asking for help 
>in understanding the 'phasing method' of single 
>sideband (SSB) demodulation.
>
>You guys helped me out, and I learned so much that 
>I wrote a blog on that SSB demodulation method.  
>In case anyone's interested, the blog is at:
>
>http://www.dsprelated.com/showarticle/176.php
>
>See Ya',
>[-Rick-]

Cool!

There is, however, an advantage to transmitting analog DSB, which is
that it works as a sort-of R = 1/2 coding.   In the demodulator the
two sidebands will be identical except for their spectral noise
contents and any asymmetric distortion that they may have received
along the way (due to the propagation channel or imperfections in the
modulator and demodulator, antenna cabling, etc.).

The two SSB halves of the DSB signal can be combined together to
reduce the noise by 3dB and potentially also help recover from some of
the channel distortions.   The carrier energy can be used as a
reference so that the two halves are exactly coherently combined.

So the power in the second sideband, or the protruding carrier, are
not wasted if used properly.  The carrier energy can also be used to
assure that the signal is translated exactly to baseband frequency so
that music (and other tonal information) is properly reproduced.
Many people can tell when music is a bit too high or too low in
frequency when reproduced.

SSB is mostly good for voice or other signals where exact frequency
recovery is not important.   Done properly the 3dB noise gain from
combining the two SSB halves of a DSB signal in the receiver is made
moot by putting all of the transmit power in the SSB, and a bit of
additional link margin is attained by suppressing the carrier energy.

Some systems do an in-between combination where some of the carrier
energy is left and the second sideband is not completely filtered
away.   This is usually referred to as "Vestigial Sideband" and the
residual carrier can be used to mix the signal back to exactly
baseband for accurate spectral registration (just like with broadcast
AM audio).   Analog TV systems worked this way and in the US and other
countries where 8-VSB is used for digital television this is also
done.

For these systems there only needs to be enough carrier protruding to
accurately phase lock the signal at baseband.  The NTSC analog TV
signals also had other subcarriers to assure proper synchronization of
color and audio information.






Eric Jacobsen
Anchor Hill Communications
www.anchorhill.com

On 9.8.12 1:50 , Eric Jacobsen wrote:
> On Wed, 08 Aug 2012 15:04:38 -0700, Rick Lyons
> <R.Lyons@_BOGUS_ieee.org> wrote:
>
>>
>> Hi guys,
>>   Some weeks ago I posted a message asking for help
>> in understanding the 'phasing method' of single
>> sideband (SSB) demodulation.
>>
>> You guys helped me out, and I learned so much that
>> I wrote a blog on that SSB demodulation method.
>> In case anyone's interested, the blog is at:
>>
>> http://www.dsprelated.com/showarticle/176.php
>>
>> See Ya',
>> [-Rick-]
>
> Cool!
>
> There is, however, an advantage to transmitting analog DSB, which is
> that it works as a sort-of R = 1/2 coding.   In the demodulator the
> two sidebands will be identical except for their spectral noise
> contents and any asymmetric distortion that they may have received
> along the way (due to the propagation channel or imperfections in the
> modulator and demodulator, antenna cabling, etc.).
>
> The two SSB halves of the DSB signal can be combined together to
> reduce the noise by 3dB and potentially also help recover from some of
> the channel distortions.   The carrier energy can be used as a
> reference so that the two halves are exactly coherently combined.
>
> SSB is mostly good for voice or other signals where exact frequency
> recovery is not important.   Done properly the 3dB noise gain from
> combining the two SSB halves of a DSB signal in the receiver is made
> moot by putting all of the transmit power in the SSB, and a bit of
> additional link margin is attained by suppressing the carrier energy.

For detecting a DSB signal, the re-inserted carrier has to be
at proper frequency and phase, otherwise severe distortion
occurs in the recovered baseband. The presence of both sidebands
makes the process sensitive to carrier phase.

(As an off-topic: Narrow-band phase modulation can be produced
combining a suppressed-carrier DSB signal with the original carrier
re-inserted at 90 degrees phase offset. The mathematics is left
as an excercise).

> So the power in the second sideband, or the protruding carrier, are
> not wasted if used properly.  The carrier energy can also be used to
> assure that the signal is translated exactly to baseband frequency so
> that music (and other tonal information) is properly reproduced.
> Many people can tell when music is a bit too high or too low in
> frequency when reproduced.

In frequency-shifted music, the harmonic relationships are broken,
and that is much easier to hear than the absolute pitch. This is
also the main cause for the Donald Duck effect, but in speech it
spoils less than in music.

-- 

Tauno Voipio

On Wednesday, August 8, 2012 6:04:38 PM UTC-4, Rick Lyons wrote:
> Hi guys,
> 
>   Some weeks ago I posted a message asking for help 
> 
> in understanding the 'phasing method' of single 
> 
> sideband (SSB) demodulation.
> 
> 
> 
> You guys helped me out, and I learned so much that 
> 
> I wrote a blog on that SSB demodulation method.  
> 
> In case anyone's interested, the blog is at:
> 
> 
> 
> http://www.dsprelated.com/showarticle/176.php
> 
> 
> 
> See Ya',
> 
> [-Rick-]

Hi Rick - great article - you put a lot of work into it. While on this subject, you may want to look at compatible single sideband modulation (CSSM). This is a SSB signal that works on standard receivers!! I have some info if you need it.

I recall cell sites comonly used gps guided clocks that were held to 0.5 parts per billion (10^9) as their frequency reference.

Clay

"Rick Lyons" <R.Lyons@_BOGUS_ieee.org> wrote in message 
news:hbo528d30hi5hi3ffs90c3htqbv4fr6dce@4ax.com...
>
> Hi guys,
>  Some weeks ago I posted a message asking for help
> in understanding the 'phasing method' of single
> sideband (SSB) demodulation.
>
> You guys helped me out, and I learned so much that
> I wrote a blog on that SSB demodulation method.
> In case anyone's interested, the blog is at:
>
> http://www.dsprelated.com/showarticle/176.php

1. When communication engineers sketch out an audio spectrum, they represent 
it as trapezium with the fall from low to high frequencies. This is more 
like the audio spectrum looks in reality, and this convention makes real 
easy to see if a sideband signal is inverted in frequency or not. The author 
draws spectrums vise versa.

2. When comparing SSB or DSB performance to FM or AM, it is often overlooked 
that the transmit power amplifier for SSB/DSB has to be linear, whereas 
FM/AM amplifiers operate in efficient class C or D or E mode. So, if 
performance is compared for the same power consumption of the transmitter 
(and this is what matters), the advantages of SSB/DSB are not so obvious if 
any at all.

3. DSB demodulator requires perfect carrier recovery in frequency and phase. 
A practical DSB demodulator must include PLL for carrier recovery. Costas 
loop structure is suitable for that purpose.

Vladimir Vassilevsky
DSP and Mixed Signal Consultant
www.abvolt.com

<clay@claysturner.com> wrote in message 
news:38f851a3-793d-49b9-a032-ac5050731c56@googlegroups.com...

> Hi Rick - great article - you put a lot of work into it. While on this 
> subject, you may want to look at compatible single sideband >modulation 
> (CSSM). This is a SSB signal that works on standard receivers!! I have 
> some info if you need it.
> I recall cell sites comonly used gps guided clocks that were held to 0.5 
> parts per billion (10^9) as their frequency reference.

Dr. Clay,
I recall your article on designing Hilbert FIRs. Very elegant and useful 
method. I've used this approach for spectrum shifting, which is essentially 
the same thing as SSB demodulator.

Vladimir Vassilevsky
DSP and Mixed Signal Consultant
www.abvolt.com

On Aug 9, 11:19&#4294967295;am, "Vladimir Vassilevsky" <nos...@nowhere.com> wrote:
> <c...@claysturner.com> wrote in message
>
> news:38f851a3-793d-49b9-a032-ac5050731c56@googlegroups.com...
>
> > Hi Rick - great article - you put a lot of work into it. While on this
> > subject, you may want to look at compatible single sideband >modulation
> > (CSSM). This is a SSB signal that works on standard receivers!! I have
> > some info if you need it.
> > I recall cell sites comonly used gps guided clocks that were held to 0.5
> > parts per billion (10^9) as their frequency reference.
>
> Dr. Clay,
> I recall your article on designing Hilbert FIRs. Very elegant and useful
> method. I've used this approach for spectrum shifting, which is essentially
> the same thing as SSB demodulator.
>
> Vladimir Vassilevsky
> DSP and Mixed Signal Consultantwww.abvolt.com

Thanks Vlad,

I'm glad you found it useful. I developed that method when I was doing
a project a while back. I find it is problems that drive the need for
solutions.

Clay

If from birth we all listened exclusively to frequency-shifted voices, do you think we would adapt to the non-harmonic harmonics and think that non-shifted voices sounded bizarre?

Bob

Robert Adams <robert.adams@analog.com> wrote:

> If from birth we all listened exclusively to frequency-shifted 
> voices, do you think we would adapt to the non-harmonic 
> harmonics and think that non-shifted voices sounded bizarre?

Someone once did an experiment generating "music" with a 
stretched (factor of 2.1) octave. As expected, it sounded
unmusical.

The problem with doing it from birth is that there are
too many natural harmonic sounds. 

Reminds me, on a lecture about color vision and the brain,
I wondered afterward if we grew up in a colorless world from
birth if we wouldn't be able to see color. That is, the optic
nerve to brain wiring for color detection wouldn't form. 

As with your question, I presume the experiment won't be done.

-- glen

Yes, that experiment will never be done. Closest thing would be a baby who spends a lot of time in Dads ham radio ssb shack :)

My question had more to do with whether the ear/brain was tuned to recognize harmonic relationships at the hardware level, or whether this is learned. In the dsp world, autocorrelation is a good way to detect periodicity even for signals with high harmonic content, but this would no longer work for frequency-shifted signals. So if the brain were using autocorrelation in some way, you would think that a frequency shifted spectrum would cause a lot of trouble.

When you think about it, it's remarkable that we can decode frequency-shifted voices at all, but in fact even with extreme frequency shifting where the harmonic relationships are way off, we can still decode what is being said.

Bob