How does an amplifier inadvertently demodulate a radio signal?

[musichascolors]

I know sometimes speakers/amps will pick up a radio signal. I'm confused as to how the signal is demodulated accidentally. Are the signals picked up exclusively AM, or do FM signals get picked up as well?

 

[berkeman]

I know sometimes speakers/amps will pick up a radio signal. I'm confused as to how the signal is demodulated accidentally. Are the signals picked up exclusively AM, or do FM signals get picked up as well?

A common way is via rectification. If you rectify an RF signal, you will generally get the envelope information, which can be audio. Certainly this can happen for AM signals, but it can happen for FM signals as well, if there is some filtering going on in the rectification path.

 

[musichascolors]

Thanks!

I'm quite new to the subject. So still trying to gain an intuitive understanding...

 

[davenn]

Thanks!

I'm quite new to the subject. So still trying to gain an intuitive understanding...

and just to expand on Berkeman's rectification comments

Two common ways this happens
1) is by the amplifier interconnect cables picking up the RF signal and the diode action of the transistors in the preamp stages
rectify and demodulate the signal which is then passed along the amplifier stages along with the wanted signal
2) is by similar process as above but when the RF is picked up by the long speaker cables the process is confined to the rectification /demodulation, of the RF signal within the output transistor stages of the amplifier

so in that last scenario, the RF is making its way into the amplifier via the speaker cables and into the
output transistors, getting rectified and amplified and sent back out to the speakersDave

 

[jim hardy]

I'm quite new to the subject. So still trying to gain an intuitive understanding...

We have no idea of your background.
Have you ever studied radio or electronics at all? Ever build a crystal set as a kid ?
Do you know what the guys meant by "rectification" ?

A picture might be worth a thousand words... and some vocabulary in order.
from http://www.radio-electronics.com/in...plitude-modulation-detection-demodulation.php

Left waveform is an AM radio signal. It is a high frequency AC wave, usually sinusoidal , and of the Radio station's Frequency . It's called RF for short.. actually "RF carrier" because it "carries" the lower frequency information that is to be conveyed.

It is "amplitude modulated" meaning its amplitude shrinks and swells as shown, at the much lower rate of the information it's carrying maybe voice, music or data.. That one looks rather like it's modulated with a sine wave, maybe it's a pure note from a violin or Madonna's pretty voice singing something.
The tips of the RF wave form the "Envelope" so called because when you look at it with an oscilloscope the individual RF waves are packed so close together they just look like a solid mass , ie they are 'enveloped'(surrounded) by the modulating signal.

" Demodulation" at its simplest is a two step process, Rectify and filter.
"Rectify" means chop off and discard half of it , in this case we've removed everything below zero volts, the negative half, leaving just the positive half of the RF wave. A simple diode will rectify the RF .
That leaves you with the middle signal. Usually the rectification process is called "Detection" and the rectifier is called a "Detector" because there are other methods of demodulation. This one is the simplest and has been around for a hundred years.

Filling in the gaps between the rectified RF peaks completes the demodulation. A simple capacitor that's not too big will accomplish this. It is sized large enough to hold charge between the RF peaks but not large enough to hold charge between the slower peaks of the envelope.
That capacitor connects the RF peaks just like connecting dots, leaving you with just the envelope frequency.
So the RF has been "filtered out" , completing the demodulation.
You're left with the third waveform.

Now - does demodulation have to be perfect ? of course not. You don't need an ideal diode,, any nonlinearity will distort the RF and partially demodulate it.

So when a wire from a microphone or speaker carries a radio signal into an amplifier,
the first transistor junction it encounters is a nonlinear device and it will partially rectify , hence imperfectly demodulate, the RF signal.
That's why the sound you hear from such accidental demodulation is not very clear. The ear and brain do a very good job of recognizing even severely distorted signals though so it's quite distracting.

A radio receiver can be this simple, note detector diode and capacitor driving headphones just as described above
read about them here http://www.hws.org.au/Crystalset%202000/mystery.html

hope this helps

if you're more advanced than i thought - no offense meant.

old jim

 

[musichascolors]

We have no idea of your background.
Have you ever studied radio or electronics at all? Ever build a crystal set as a kid ?
Do you know what the guys meant by "rectification" ?

I know what a rectifier does in terms of converting to one polarity, but I don't yet understand it's relevance to this application.

I'm fairly knowledgeable in terms of wave physics, overtones, modulation, etc (just from working with audio), but know very very little about electronics and circuits. So an explanation based in wave physics would be very helpful.

Edit: Thank you for your explanation. I'm starting to get it, but I don't yet understand some things. My understanding (possibly incorrect) is that the audio signal is just summed onto the carrier frequency (or is that not the case?) If so, why does the RF frequency need to be filtered, wouldn't it be inaudible anyway? And why is the rectification necessary (why do we have to remove everything below 0 volts)?

Thank you!

 

[jim hardy]

My understanding (possibly incorrect) is that the audio signal is just summed onto the carrier frequency (or is that not the case?)

I think mathematically Amplitude Modulation is a multiplication not addition...

If so, why does the RF frequency need to be filtered, wouldn't it be inaudible anyway?

Remember, 100 years ago the intent was to recover voice or operate an electromagnet for telegraphy.
In that crystal set the inductance of the headphones would not pass RF , X_L = 2πfL
in communication radios the detected output is handed to audio frequency amplifiers for delivery to a loud speaker.
So removing the RF allows simpler audio amplifiers. Still done today.

And why is the rectification necessary (why do we have to remove everything below 0 volts)?

There are other methods, see http://web.mit.edu/6.02/www/s2012/handouts/14.pdf
but the answer to your question is it's the easiest way to demodulate.
in the early days of radio it was easy to find crystals or metal oxides that performed rectification , So that's what they did.
In grade school i had a crystal set.
in high school (early 1960's ) when i was taught radio basics we used vacuum tube rectifiers .

Think of it like this -
Look at left waveform in post #5.
What is the short term average of the last few RF cycles ? It's zero because the average of a symmetric wave is zero.

Now look at the middle waveform in post #5.
What is the short term average of the last few rectified RF cycles? It's a positive value that varies at the rate of the signal we wished to convey. How very convenient !
Mathematically, a short term averager is a low pass filter.

Sorry, I'm not well versed in high math or wave functions, i barely passed vector calculus 50 years ago and know better than to tangle with it now.

Any help ?

old jim

 

[musichascolors]

Thanks, the multiplication bit is helpful in trying to get an intuitive understanding.

Based on my username, you can probably parse that I'm coming at this from a musical background. So maybe using this scenario would be helpful (for me)?

So if, for the sake of my understanding... If I had a carrier frequency of 10 khz, and was trying to, using AM, transmit a sine wave of 4khz, how would I do that (would I even be able to?). My understanding is that just summing the two waves would cause the amplitude of the 10khz carrier to be modulated up and down at a rate of 4khz, but is that not what happens?

EDIT: Or I guess the modulation is more like having a tremolo effect at the modulating frequency, which would obviously produce different results...?

 

[mheslep]

A common way is via rectification. If you rectify an RF signal, ...

Rectification, and low pass filtering, the filtering provided by the bandwidth response of the amplifier.

 

[jim hardy]

Glad you mentioned tremolo. I was about to flub up a mathematical attempt.
Nowadays there's a field of EE called "Digital Signal Processing" , DSP.
Have you studied Fourier Analysis?
DSP uses it intensely. I don't go there much.

Addition and multiplication of sinewaves don't give the same result.
Addition just gives sin(a) + sin(b), the original frequencies.
multiplication gives

and observe taking the product gives you two new frequencies, sum and difference of the originals.
Those are called the "sidebands" because they are centered on the mean and differ by twice the modulating frequency.
That mean is the carrier, sum and difference are the sidebands. They sum to that enveloped wave in post #5.

To help visualize the difference in sum and product
i figure this picture saves a thousand words
it's a picture of sinewave addition
from http://clas.mq.edu.au/speech/acoustics/waveforms/adding_waveforms.html
that i annotated

multiplication of any two numbers has to give a zero result whenever either one is zero, doesn't it ? as in leftmost wave of post 5 ? Here you can SEE that isn't the case for addition.
So addition of two isn't amplitude modulation. It's addition.

When you want to Amplitude Modulate a sinewave, you multiply it by another one.
That's not too difficult with tubes or transistors
but nowadays everybody uses specialized integrated circuits
(well radio stations probably still use giant tubes, maybe our Ham radio friends on PF will elucidate)
from http://www.analog.com/media/en/technical-documentation/data-sheets/AD633.pdf

Note the multiplier gives just the sidebands, in agreement with the trig identity up above for products, and they add the carrier back in at pin 6. That's why i like electonics , it is so close to math.
But unlike normal people i have to use electronics to understand math not vice versa.

Last time i tried to derive ths from trig identities i had to be rescued by a kindly PF member whose math is better than mine.

I hope this helps you. If you need to see the the DSP math, i'd suggest post a question in the math forum...
sinewaves are fascinating. Euler was a genius.old jim

 

[jim hardy]

How a real analog multiplier works

It is easy to make a controlled current source using transistors
and it is easy to make another transistor behave like an adjustable resistor
so place them in series and by ohm's law and you have multiplied .

that's called a "transconductance " multiplier.

There are other types.

 

[Averagesupernova]

Why don't you just do something up in a spreadsheet and graph it or write a quick program? You can get a good feel for what is happening by doing that. AM is most certainly multiplication of the modulating signal and the carrier.

 

[musichascolors]

Thank you for clarifying the multiplication bit.

I guess I was trying to understand it in a less mathematical way. For example my understanding of FM is that you basically convert the signal to a "vibrato" which then modulates the carrier frequency. The detector then outputs 0v when the carrier frequency is at exact pitch and the voltage goes up and down depending on the degree of pitch change, so the signal gets recreated. I probably wrote that poorly, but it seems like kind of an intuitive way of understanding how FM works, and I was hoping to get a similar level of understanding for AM.

But in this case, maybe just thinking mathematically is more straightforward.

 

[jim hardy]

i don't know much about music, looked up vibrato and it seems the right conceptYour understanding of FM is exactly right.
FM broadcast in US(your 88 to 108 mhz radio stations) shifts the carrier frequency ±150khz for full modulation.
That'd be like modulating the pitch of a sound.
FM was used also to record analog data in scientific investigations before the computer based dataloggers came into existence. An audio tone is frequency modulated per the measurement and recorded on an old fashioned tape recorder.
I had an Ampex 32 channel FM tape recorder in the power plant, allegedly portable it weighed a hundred pounds but was a monument to precision. AM of course adjusts the amplitude instead of the frequency
that same site i linked earlier has more on modulation
http://www.radio-electronics.com/in...-amplitude-modulation/what-is-am-tutorial.php

Amplitude modulation basics
When an amplitude modulated signal is created, the amplitude of the signal is varied in line with the variations in intensity of the sound wave. In this way the overall amplitude or envelope of the carrier is modulated to carry the audio signal. Here the envelope of the carrier can be seen to change in line with the modulating signal.

Suoernova had a good idea above-
Are you good with spreadsheets ?
Try setting a cell for sin(t) and another for sin(t/50)
t in degrees not radians
and a third cell for their product cell1 X cell2

run t from 0 to 36000 and plot third cell
it should plot 100 cycles of amplitude modulated carrier showing 2 cycles of envelope, resembling the picture above.

then run it again showing sum instead of product

i always did mine in Qbasic... xcel only taunts me .

old jim

 

[Averagesupernova]

To confuse you even farther, the carrier frequency itself does not change in amplitude with AM. It appears so but it is not the case. The composite signal of the carrier and the upper and lower sidebands make it appear like the carrier itself is changing but it is really not.

 

[musichascolors]

So is tremolo not a good analogy in that case, if the carrier frequency isn't changing in amplitude?

 

[jim hardy]

The Fourier transfer term for carrier doesn't change amplitude

but as you can see in the pictures all RF cycles do not remain the same height.
Algebraically the carrier and modulation are multiplied.
But in the Fourier series of the result , sideband terms are added.

Fourier and Laplace transforms are tools that i learned to use for cracking textbook problems, without understanding their underlying math.
Just as a crow learns to drop a clam from altitude to crack it open, without understanding the math of momentum and impulse.

So is tremolo not a good analogy in that case, if the carrier frequency isn't changing in amplitude?

I looked up Tremolo in Webster and found it refers to either pitch or loudness.
So i left that one lone.

old jim

 

[musichascolors]

I'm familiar with Fourier transforms in the sense that I know what they are and (in some cases) what they're used for. And just generally looking at spectral analyzers, looking at waveforms, playing with tools/toys like this http://www.falstad.com/fourier/

So I'm quite familiar with Fourier transforms, but in more of the practical/applied sense than in the math formula sense.

The Fourier transfer term for carrier doesn't change amplitude
I looked up Tremolo in Webster and found it refers to either pitch or loudness.
So i left that one lone.

old jim

Pitch is incorrect, tremolo and vibrato are often used interchangeably but they're not the same thing. Pitch(vibrato) would be frequency modulation and volume (tremolo) would be amplitude modulation. That's why I thought it would be applicable(?)

 

[musichascolors]

When I have a bit of free time, I'm going to see if I can mess around with my synthesizer/recording software to try and see if I can encode/decode AM that way. Would help me understand it intuitively.

 

[mheslep]

Matlab seems to be software of choice at the moment for communications work, at least in academia. One can generate a signal, perform any kind of modulation, characterize a channel, dump statistics, and plot resulrs all in a few dozen lines of code.

 

[jim hardy]

Pitch is incorrect, tremolo and vibrato are often used interchangeably but they're not the same thing. Pitch(vibrato) would be frequency modulation and volume (tremolo) would be amplitude modulation. That's why I thought it would be applicable(?)

in that case you have it right.

 

[jim hardy]

Matlab seems to be software of choice at the moment for communications work, at least in academia. One can generate a signal, perform any kind of modulation, characterize a channel, dump statistics, and plot resulrs all in a few dozen lines of code.

is it hard to learn ? Excel was nothing but frustration and led to serial mouse-icide.

 

[mheslep]

is it hard to learn ? Excel was nothing but frustration and led to serial mouse-icide.

Easy to learn and powerful, that's not the drawback. Matlab is proprietary, costs few hundred $ for non-students, more for some of the plugin packages, which often causes problems I find as soon as one wants to ship code around to others outside your own shop. Also, its not quite a proper programming language, so as soon as a problem with some complexity and size comes along, requiring many pieces to cooperate, or requires high, C type compute performance, or something that requires longer term maintenance, Matlab becomes troublesome. For these reasons, the last serious team project I worked used Matlab to prototype all the technically challenging algorithms, and then converted the most successful Matlab algorithms to C++.

http://www.mathworks.com/help/comm/examples/llr-vs-hard-decision-demodulation.html

 

[musichascolors]

Playing around with my synthesizer, I am able to see the sidebands develop as I increase the rate of modulation (up to a point)

The only clumsy part is I'm having to do it with a triangle wave, which is the closest I can get to a Sin.

But I can still see the sidebands develop around the fundamental of the triangle.

 

[jim hardy]

A pretty good writeup here

http://www.egr.msu.edu/em/research/goali/notes/module7_am_fm.pdf

 

[CWatters]

Re: How does an amplifier inadvertently demodulate a radio signal?

During WW2 the landlord of a British public house called to police to report hearing "someone speaking German up my chimney". Turned out they weren't far from a BBC transmitter used to broadcast messages to Europe, including some in German. As I heard it the chimney was metal lined and somehow a joint partially rectified the signal causing the transmissions to become audible.

 

[dlgoff]

If you like animations, http://www.animations.physics.unsw.edu.au/ have some cool ones on sound. e.g. Quantifying sound

 

[tech99]

I know sometimes speakers/amps will pick up a radio signal. I'm confused as to how the signal is demodulated accidentally. Are the signals picked up exclusively AM, or do FM signals get picked up as well?

As a matter of interest there are two main ways that an amplifier device (such as a transistor or vacuum tube) can cause rectification and demodulation of an AM signal. Assuming a bipolar transistor:-
1) The usual input connection between base and emitter is a diode, and unless the characteristic is swamped out with a high resistance, it will rectify. For a tube, this rectification is used for the "leaky grid detector".
2) The input/output characteristic (the mutual characteristic) is usually curved towards the upper and lower ends, as overload is approached. If the bias sets the device near a curved part, then we have "collector bend" or "anode bend" rectification.

 

[musichascolors]

So this seems to be an interesting demo

I converted into mp3 and then ran it through a software program that rectified it. Then I lowpassed it so frequencies above 500 hz were removed (the carrier frequency is 558)

It did indeed sound smoother.

But why does the rectification lower the distortion?

 

[tech99]

Not sure what you have done, but the carrier needs to be twice as high as the maximum audio, otherwise the sidebands and audio get mixed up together.

 

[musichascolors]

Sorry, here's the demo



"I converted into mp3 and then ran it through a software program that rectified it. Then I lowpassed it so frequencies above 500 hz were removed (the carrier frequency is 558)"

Ok so i'll do another low pass at 225 hz (or 200 to be safe)

there won't be much audio in that case but yeah.

Also your comment is really helpful, thanks!

 

[jim hardy]

That doesn't look rectified...

 

[musichascolors]

Well I don't think it's rectified in the audio example

 

[AlexCaledin]

During WW2 the landlord of a British public house called to police to report hearing "someone speaking German up my chimney". Turned out they weren't far from a BBC transmitter used to broadcast messages to Europe, including some in German. As I heard it the chimney was metal lined and somehow a joint partially rectified the signal causing the transmissions to become audible.

More likely, the strong RF current created a magnetic force which's proportional to the square of current, thus having the audio frequency and turning that chimney into a sort of speaker...
As for a joint, it could make RF sparks also creating quadratic force.

 

[Averagesupernova]

More likely, the strong RF current created a magnetic force which's proportional to the square of current, thus having the audio frequency and turning that chimney into a sort of speaker...
As for a joint, it could make RF sparks also creating quadratic force.

You won't hear it until it is detected which was likely done by the joint in the chimney. If all it takes is strong RF then every conductor at every transmitter site would be blasting with some sort of audio.

 

[jim hardy]

early rectifiers were copper oxide... my guess would be a joint in the roof flashing somewhere

 

[Baluncore]

To demodulate a signal requires a non-linear electrical junction such as a semiconductor.

There are many natural semiconductors available in an amplifier's environment. For example; zinc oxide, copper oxide and carborundum grinding grit. Any signal connector contaminated with a semiconductor particle can play a part in the demodulation.

 

[Baluncore]

During WW2 the landlord of a British public house called to police to report hearing "someone speaking German up my chimney". Turned out they weren't far from a BBC transmitter used to broadcast messages to Europe, including some in German. As I heard it the chimney was metal lined and somehow a joint partially rectified the signal causing the transmissions to become audible.

Which reminded me of another semiconductor, discovered in, and named after Germany.
Germanium was concentrated by plants that then become coal. The coal is compressed over time then, when the coal is burned, the germanium vapour condensed on the walls of the refinery or power station flue. One of the original sources of germanium for semiconductors was from the chimney stacks of German furnaces. So a coal fire below a metal lined chimney would be quite capable of condensing the semiconductor crystal needed to demodulate the local signal.

 

[sophiecentaur]

So is tremolo not a good analogy in that case, if the carrier frequency isn't changing in amplitude?

Tremolo is a fair way to describe FM as the carrier frequency is varied by the modulating signal. When the frequency deviation is high the carrier can actually be seen to sweep from side to side when you look with a spectrum analyser . That would be, for instance if you swept the carrier frequency from +10kHZ to -10kHz relative to the steady value (of 100kHz) at a rate of, say 50Hz. But sometimes the frequency deviation is low and the carrier deviation might only be +/-1kHz with a modulation frequency of 2kHz. Then the spectrum of the modulated signal really becomes unrecognisable to the eye. Nonetheless, the system works and a suitable FM demodulator can dig the modulating signal out of the received signal. The Maths of FM is a nightmare compared with AM.
On the whole, though, analogies are not very fruitful and you need to use Maths in order to get a good grip of signalling processes. It is a good thing that AM, at least, can be done in pictures!

 

[jim hardy]

On the whole, though, analogies are not very fruitful and you need to use Maths in order to get a good grip of signalling processes. It is a good thing that AM, at least, can be done in pictures!

What latched in FM for me was tinkering with Phase Locked Loops (PLL's).
LM567 gives a DC output proportional to how far off center frequency it must operate to maintain phase lock.
So it'll 'de-frequency-modulate' with ease.
I once built a PLL gizmo that modulated two ultrasonic carriers , 22 and 28 khz, with telephone "TouchTone" DTMF audio tones and added it to the plant PA system to identify which of the PA stations was active.

This is a great learner's textbook on that fifty year old technology.
https://ia801608.us.archive.org/30/...s_5800304/1972_Signetics_PLL_Applications.pdf

old jim

 

[mheslep]

To demodulate a signal requires a non-linear electrical junction such as a semiconductor...

I think you mean mixing a signal (frequency shifting) requires a nonlinear element. Demodulation can be done in some cases with a simple linear filter.

 

[sophiecentaur]

So it'll 'de-frequency-modulate' with ease.

There are so many ways of demodulating an FM signal. If you off tune an AM receiver, you can 'slope detect' an fm signal by the AM detector producing an output which is proportional to the carrier frequency over the sloping response of the edge of the IF filter. That is an actual FM demodulator and there are a lot of such designs. Then there is the phase detector - which is what your PLL feedback signal is actually doing. They all have their advantages and disadvantages and the arrival of Integrated Circuits (tarrraaa!) caused a flurry of designs.

 

[Baluncore]

Demodulation can be done in some cases with a simple linear filter.

How do you demodulate AM or FM with a linear filter?

 

[sophiecentaur]

How do you demodulate AM or FM with a linear filter?

You will always need an amplitude detector of some sort as well as the linear filter. So you still need a non-linear element.

 

[jim hardy]

I learned FM limiter-discriminator in high school
i think it was Foster-Seely with vacuum tube diodes
http://www.radio-electronics.com/in...n/foster-seeley-fm-detector-discriminator.php
but have now forgot the details
Never forget aligning them though !

 

[mheslep]

How do you demodulate AM or FM with a linear filter?

You will always need an amplitude detector of some sort as well as the linear filter. So you still need a non-linear element.

Yes, sorry, a detector is required.

 

[musichascolors]

Tremolo is a fair way to describe FM as the carrier frequency is varied by the modulating signal.

Musically speaking, tremolo is when the amplitude is modulated, vibrato is when the pitch is modulated, that's why I thought they might be good analogies.[QUOTE="sophiecentaur, post: 5470557, member: 199289"On the whole, though, analogies are not very fruitful and you need to use Maths in order to get a good grip of signalling processes. [/QUOTE]

I guess it depends on what one considers a good grip. I feel that I understand things better when I actually make them work and can observe them. Going to play around some more with tremolo/rectifier plugins when i get the chance. (Though the tremolo plugins I've found only go up to about 7 hz or so, so it makes it a bit challenging.

 

[sophiecentaur]

I guess it depends on what one considers a good grip.

I think that's an easy one. A 'good grip' means the ability to predict, accurately, what will happen in any particular circumstance. Very hard to do without Maths in pretty well any form of modulation except basic AM.
(But the point is well made about the difference between tremolo and vibrato. I think that true tremolo is probably only obtainable on an electronic processor / instrument.)

 

[jim hardy]

Musically speaking, tremolo is when the amplitude is modulated, vibrato is when the pitch is modulated,

Tremble vs Vibrate ? I'm having a hard time tying one word to pitch and the other to intensity, Is there a memory peg you musical folks use?

 

[sophiecentaur]

Vibrating your finger on a violin finger board will vary the tension and length of the string. That will be FM.
Tremolo is hard to produce on wind or string without pulling the frequency too. Perhaps a vibrating shutter on the air supply to a pipe organ?