Is it possible to build this demodulator?

[KerimF]

Hi,

Let us say I need to demodulate an AM signal but its carrier component is suppressed (called DSB-SC).

For this, let us take a simple example.
The input signal (2V maximum peak, from an AM IF stage output, 455 KHz) could be written:

Vin = 2 * cos(2*PI*455e3*t) * f(t)
where f(t) is an audio signal with zero average voltage, say from 100 to 8000 Hz.

As circuit components, we can use CD4046 (PLL) and LM339 (comparator) besides various standard gates, resistors and capacitors (no coils for the demodulator).

The question is (But I guess it has to be addressed to experts in communications):

Could we build a practical DSB-SC demodulator using these (low cost and simple) components to restore properly the audio signal?

Sorry. As you will see the question is tricky because I am afraid I am the only one in the world to know how to do it (apparently in the least, as a search on the internet shows)!

Of course, there are many solutions to build a DSB-SC demodulator but the known ones are rather not practical (having high sensitivity to various factors) or complex and costly using the old method known as Costas Loop which was finally possible to implement in digital systems.

If in doubt (now this is addressed to graduate students), try asking your teachers:

Which is simpler to detect, a SSB-SC signal or a DSB-SC?

The answer will be the former (SSB-SC) though DSB-SC takes twice the bandwidth and provides the baseband on two bands!!! Yes, it sounds like a dilemma (since it contradicts the common sense; having more info gives trouble!). But this dilemma has to exist as being true because the world still ignores the very simple method to deal with DSB-SC (which can be done as analogue circuit and digital as well).

Finally and as many of you have already guessed, the purpose I am writing this, is to find someone who can do me a favour by showing me I am wrong in a way or another.

Cheers,

Kerim

 

[FvM]

As you are suggesting (although not exactly stating, in my opinion) the basic problem of SSB or DSB demodulation is to recover the carrier (frequency and phase). Unfortunately, the analog modulated DSB signal doesn't contains an unequivocal carrier phase information. Without additional means, e.g. a pilot tone, there's no simple method to regenerate the carrier. As far as I see, there are no technical applications for analog DSB-SC that work without similar means. Digital modulation has the mentioned costas loop.

 

[KerimF]

As you are suggesting (although not exactly stating, in my opinion) the basic problem of SSB or DSB demodulation is to recover the carrier (frequency and phase). Unfortunately, the analog modulated DSB signal doesn't contains an unequivocal carrier phase information. Without additional means, e.g. a pilot tone, there's no simple method to regenerate the carrier.

You are right, this is exactly what I meant.
I did build and even used in practical links (private ones) the simple DSB-SC demodulator using a conventional PLL though modified a little bit to restore the frequency and phase from the two bands only (since the carrier is suppressed). Even if the suppressed carrier is shifted, the loop still in lock in a good range (10% in the least).
To my big surprise, I found out that no university in the world knows about what I did!!!

Edited:
Obviously it is hard for any great professor in communications to ask me... how I did it :wink:
And who work for a great semiconductors company have no interest to compete their own products, in which they have already invested a lot, by producing simpler solutions which are low cost hence low profit as well :wink:
Even Science... Pride and Money seem to have the last word in its progress :shock:

 

[FvM]

I did build and even used in practical links (private ones) the simple DSB-SC demodulator using a conventional PLL though modified a little bit to restore the frequency and phase from the two bands only (since the carrier is suppressed). Even if the suppressed carrier is shifted, the loop still in lock in a good range (10% in the least).

I didn't further investigate the problem with real audio signals, but I presume, that the lock behaviour will depend on the base band frequency spectrum. If you allow the signal phase to lock arbitrarily to 0 or 180°, the method can probably work with some additional prerequisites, e.g. no extended signal idle phases.

I'm not aware of the complete literature and can't comment your impression, that nobody seems to have ever thought about it. I doubt, if there would be obvious applications for analog DSB-SC besides the well-known cases using supplemental techniques (e.g. Stereo broadcast).

 

[KerimF]

I didn't further investigate the problem with real audio signals, but I presume, that the lock behaviour will depend on the base band frequency spectrum. If you allow the signal phase to lock arbitrarily to 0 or 180°, the method can probably work with some additional prerequisites, e.g. no extended signal idle phases.

I'm not aware of the complete literature and can't comment your impression, that nobody seems to have ever thought about it. I doubt, if there would be obvious applications for analog DSB-SC besides the well-known cases using supplemental techniques (e.g. Stereo broadcast).

You are right again, for audio signal the polarity of the lock is not important as when sending binary data for example (in this case a sort of initialization is required for each packet).

On the other hand, as I said, I did use it in a practical audio links since its settling time is rather short for a normal speech and even music (it locks much like if the carrier does exist).

For instance, one of the good advantages of the DSB-SC system is that no power is transmitted during silent periods which is very important if used in the aviation links since this will reduce, to a great extent, the interference in case many planes transmit on the same band in an area (it is unlikely that all captains keep talking all the time without breaks). But also FCC has no idea how simple and reliable a DSB-SC demodulator can be, to the point they see lately that DSB-SC has no reason to exist anymore :shock:

 

[davenn]

All local to airports plane to tower comms is still plain VHF AM. It has often been thought about to go to VHF FM, but for various reasons AM is still the prefered mode worldwide. Once farther out SSB HF <20MHz is more normally used

SSB is still preferable to DSB primarily cuz of the narrower bandwidth, of which one of the main advantages is better S/N ratio

Dave

 

[KerimF]

All local to airports plane to tower comms is still plain VHF AM. It has often been thought about to go to VHF FM, but for various reasons AM is still the prefered mode worldwide. Once farther out SSB HF <20MHz is more normally used

SSB is still preferable to DSB primarily cuz of the narrower bandwidth, of which one of the main advantages is better S/N ratio

Dave

You are right and I agree with you about all what you said.

On the other hand, I have noticed from the many comments I got, since about 6 months, that there is a common point in all of them which says (by an analogy):
"Since we have developped supper fast adders why to waste time and learn how to do a multiplication".

I mean it becomes normal to teach the students that DSB-SC is rather hard to demodulate (hence not recommended) though this contradicts the common sense; receiving more info makes detection harder! This means, to me in the least, that modern science doesn't need to be based on logic anymore... Am I wrong? I was surprised for months that no one in the scientific community tried, in a way or another, to put an end to this false belief, for himself in the least.

For instance, what I didn't say yet is that I did this simple DSB-SC demodulator (which works with any amplitude modulation index as well) in 1979 as an MS thesis but I couldn't submit it since I had no support at that time and I needed to start my private business as soon as possible to gain my living (by producing electronic circuits for the local market demands). But later, I have always thought that what I did, being simple and good, many other engineers will have, in a way or another, the chance to do it as well. But the dilemma which I used to hear at the university and for which I decided to choose the subject of my final thesis, does exist till these days!

 

[FvM]

I don't think, that the discussed DSB-SC would be a candidate to replace SSB. SSB is used in cases, where you accept the lack of an exact carrier for demodulation and involved frequency shifts. Technically, SSB and analog modulation in general is legacy, but will accompany us for decades anyway. New analog standards will hardly be established, however.

But apart from application chances, the technical properties of the suggested analog DSB-SC with carrier recovery can be investigated in a pure academic discussion. I feel discomfortable with a technique, that's only exactly guaranteed to lock under good conditions. At least, the behaviour in presence of noise should be analyzed. When you claim an advantage for "no power is transmitted during silent periods", than you have to check the case, that a signal level is slowly rising in presence of noise. You'll need a squelch, which means, that the technique is only applicable to low quality speech signals.

 

[KerimF]

I don't think, that the discussed DSB-SC would be a candidate to replace SSB. SSB is used in cases, where you accept the lack of an exact carrier for demodulation and involved frequency shifts. Technically, SSB and analog modulation in general is legacy, but will accompany us for decades anyway. New analog standards will hardly be established, however.

But apart from application chances, the technical properties of the suggested analog DSB-SC with carrier recovery can be investigated in a pure academic discussion. I feel discomfortable with a technique, that's only exactly guaranteed to lock under good conditions. At least, the behaviour in presence of noise should be analyzed. When you claim an advantage for "no power is transmitted during silent periods", than you have to check the case, that a signal level is slowly rising in presence of noise. You'll need a squelch, which means, that the technique is only applicable to low quality speech signals.

I like to add that, decades ago, I used this technique on the MW band as a secure private audio link. I knew that there was no receiver being able to demodulate clearly a DSB-SC transmitted signal, at that time. And to make the link less noticed by listeners, I, intentionally, let the frequency of the suppressed carrier to oscillate in a rather wide range, about +/- 50 KHz, though at a low rate (around 6 Hz if I remember well). Ordinary receivers gave a sort of periodic noise (irregular in amplitude) on their loudspeakers as if it comes from a sort of electrical machines that keep running and stopping. But on our receivers, the conversation was always clear as for normal AM broadcast.
My other attempt was building a link on the FM band. I let the low baseband be empty (known as L+R), and I modulated the voice (200Hz to 5000Hz) on 32768 Hz (since I had its crystal) as DSB-SC. Obviously there was no need for any pilot. The ordinary FM receiver shows an empty/silent channel while on my side I was able to restore the audio signal from the ultrasonic signal using the same concept.
I mean, to me, it wasn't just an academic project but an applied one as well (and for many years). Obviously the concept can be made in digital form as anything else when necessary (after all it is much simpler than Costas Loop).

Edited:
Just an added note. While searching on the subject, I found out that a solution that I considered as a failure while I was working on my thesis, happened to have a patent! it is about reversing the AM envelop at zero-crossing though the original amplitude may sometimes be zero without changing its sign afterward.

Edited: I re-studied the last idea (zero-crossing). It may work but detecting the state of the carrier phase near zero amplitude may take time, mainly in the presence of noise, before deciding if the sign of the envelop needs to be changed or not.