How did Armstrong's FM receiver and transmitter work?

[Cool4Kat]

Hey, I was wondering if any of you smart people could help explain in a very simple way (if possible) how the first FM transmitter and receiver worked?

I know, for example, that many current transmitters use a voltage variable conductor to change the frequency of the wave. Did Armstrong do that in 1933? I have tried to read his patents but I don't get it.

Also, did he use a superheterodyne method or a heterodyne method to receive the FM (or some other method)?

Thanks so much!

 

[tech99]

For the transmitter, Armstrong started with a low frequency in the kHz region. He used a phase modulator, not a frequency modulator. If you look at the phasor diagram of PM compared with AM, the only difference is that the carrier for PM is 90 degrees phase shifted. So Armstrong used a balanced modulator to first obtain two AM sidebands with no carrier. Then he inserted a new carrier 90 degree phase shifted from the original. This was achieved at low frequency, and he used a long chain of frequency multipliers to obtain the final frequency at VHF. The frequency multipliers used tube amplifiers biased well below cut-off to obtain harmonics.
PM and FM are very similar and we can convert one to the other by simple audio correction. Armstrong's phase modulator produced only very slight FM, but the long chain of frequency multipliers increased the frequency deviation to the required amount.
In the receiver, I believe Armstrong used a resonant LC circuit set to one side of the signal, so that variations in frequency produced a varying voltage. He used a superheterodyne and the demodulation process happened at intermediate frequency. Prior to demodulation, the signal was amplified very greatly and passed through limiter, or clipping, stages to remove unwanted AM signals such as ignition interference.

 

[Averagesupernova]

In the receiver, I believe Armstrong used a resonant LC circuit set to one side of the signal, so that variations in frequency produced a varying voltage.

I believe this is referred to as slope detection and while I don't know for fact that this is what Armstrong used it is a pretty safe bet.

 

[tech99]

I have found a website showing an Armstrong receiver used for technical monitoring. It seems to have a centre-zero meter on it, which tends to indicate that it had more than a simple slope detector. As the Foster-Seeley detector had not been invented, it may be that Armstrong used two tuned circuits, one above and one below the carrier. This type was known as the Round-Travis circuit in the UK in the 1960s, where it was used for FDM/FM microwave links.
http://aggregate.org/hankd/REL/

 

[Cool4Kat]

For the transmitter, Armstrong started with a low frequency in the kHz region. He used a phase modulator, not a frequency modulator. If you look at the phasor diagram of PM compared with AM, the only difference is that the carrier for PM is 90 degrees phase shifted. So Armstrong used a balanced modulator to first obtain two AM sidebands with no carrier. Then he inserted a new carrier 90 degree phase shifted from the original. This was achieved at low frequency, and he used a long chain of frequency multipliers to obtain the final frequency at VHF. The frequency multipliers used tube amplifiers biased well below cut-off to obtain harmonics.
PM and FM are very similar and we can convert one to the other by simple audio correction. Armstrong's phase modulator produced only very slight FM, but the long chain of frequency multipliers increased the frequency deviation to the required amount.
In the receiver, I believe Armstrong used a resonant LC circuit set to one side of the signal, so that variations in frequency produced a varying voltage. He used a superheterodyne and the demodulation process happened at intermediate frequency. Prior to demodulation, the signal was amplified very greatly and passed through limiter, or clipping, stages to remove unwanted AM signals such as ignition interference.

Thanks for trying to help me - I really appreciate it. However, I am not an engineer I am only a physics person and I am sorry to say I am still confused. What is a balanced modulator? What are two AM sidebands? How do you convert PM to FM with audio correction?

I seem to be stuck between sites that don't describe anything and ones that use terms I don't understand.

Thanks again,

Kathy

 

[davenn]

What is a balanced modulator? What are two AM sidebands?

Kathy

these are terms you would find the answer to in a few minutes of some personal research on google

On PF we like to see people that help themselves
do some searching and then come back with any questions and links to anything you didn't quite understandDave

 

[sophiecentaur]

Thanks for trying to help me - I really appreciate it. However, I am not an engineer I am only a physics person and I am sorry to say I am still confused. What is a balanced modulator? What are two AM sidebands? How do you convert PM to FM with audio correction?

I seem to be stuck between sites that don't describe anything and ones that use terms I don't understand.

Thanks again,

Kathy

Your problem here seems to me that you want to know too much all at once. If you have as little idea about the topic as you claim then you won't get anywhere by asking a string of questions about it. You need to read a textbook with a proper program of study and not rely on random Google hits to help you.
The stuff you want to know comes at the end of a basic course on electronics and signal processing. Sorry but I'm only saying how it is; Q and A may have been ok for ancient Greeks and their one-to-one tutors but not for EE. I reckon you would probably have the same problem if a Biologist asked you to tell them about the Zeeman effect when they knew nothing about the very basic stuff on QM.

 

[Cool4Kat]

Your problem here seems to me that you want to know too much all at once. If you have as little idea about the topic as you claim then you won't get anywhere by asking a string of questions about it. You need to read a textbook with a proper program of study and not rely on random Google hits to help you.
The stuff you want to know comes at the end of a basic course on electronics and signal processing. Sorry but I'm only saying how it is; Q and A may have been ok for ancient Greeks and their one-to-one tutors but not for EE. I reckon you would probably have the same problem if a Biologist asked you to tell them about the Zeeman effect when they knew nothing about the very basic stuff on QM.

Well OK I can only understand radios if I take a class in EE? I will keep on searching and trying to understand though and I do appreciate random people trying to help me. I would be happy to explain the Zeeman effect to a biologist, by the way!

 

[sophiecentaur]

Well OK I can only understand radios if I take a class in EE? I will keep on searching and trying to understand though and I do appreciate random people trying to help me. I would be happy to explain the Zeeman effect to a biologist, by the way!

I wonder if the average Biologist would grasp the significance of quantum numbers, magnetic dipole moment and precession? I think you could probably find yourself on a slippery slope.
If your well informed Biologist friends don't fit my analogy then we could easily find another. I still feel that you claim to know so little that you need more basics.

What is a balanced modulator? What are two AM sidebands? How do you convert PM to FM with audio correction?

Those three ideas deserve threads of their own and you can bet your bottom dollar that each thread would introduce a few other new terms. I am sure you didn't get to your level in Physics using a thread on a forum and that is the problem. Perhaps your choice of an ancient patent was not the best way into a subject. Patents are not written in an educational way; they are written to present the assessors with enough information to persuade them that the applicant's idea is defined in a way that includes what makes the idea special / novel and the information in a good patent should make obvious if someone tries to copy it in a sneaky way. (No one actually checks whether or not the invention will work) If you want a way into this then why not try a textbook?
"Well OK I can only understand radios if I take a class in EE? " I am not questioning your inherent ability or potential as an Electronics Boffin but I would say it would need to be that or a long time of enthusiastic involvement with a Radio Ham group or extended playing with construction kits and reading loads of 'mags'. If you know your Physics, you will appreciate that the Maths is important. Fourier Frequency Analysis may be familiar to you and, if it is, things will be much easier for you. But EE is chock full of buzzwords and phrases that tend to be bandied about in an 'approximate fashion' in many forum (not PF) posts. You have to navigate around that sort of thing.
Wikipedia can sometimes help.

 

[Cool4Kat]

I wonder if the average Biologist would grasp the significance of quantum numbers, magnetic dipole moment and precession? I think you could probably find yourself on a slippery slope.
If your well informed Biologist friends don't fit my analogy then we could easily find another. I still feel that you claim to know so little that you need more basics.

Those three ideas deserve threads of their own and you can bet your bottom dollar that each thread would introduce a few other new terms. I am sure you didn't get to your level in Physics using a thread on a forum and that is the problem. Perhaps your choice of an ancient patent was not the best way into a subject. Patents are not written in an educational way; they are written to present the assessors with enough information to persuade them that the applicant's idea is defined in a way that includes what makes the idea special / novel and the information in a good patent should make obvious if someone tries to copy it in a sneaky way. (No one actually checks whether or not the invention will work) If you want a way into this then why not try a textbook?
"Well OK I can only understand radios if I take a class in EE? " I am not questioning your inherent ability or potential as an Electronics Boffin but I would say it would need to be that or a long time of enthusiastic involvement with a Radio Ham group or extended playing with construction kits and reading loads of 'mags'. If you know your Physics, you will appreciate that the Maths is important. Fourier Frequency Analysis may be familiar to you and, if it is, things will be much easier for you. But EE is chock full of buzzwords and phrases that tend to be bandied about in an 'approximate fashion' in many forum (not PF) posts. You have to navigate around that sort of thing.
Wikipedia can sometimes help.

All I am looking for is a basic explanation of how he made the frequency change. For example, in another thread I saw that they use variable capacitors to change the capacitance of a LC circuit to make FM waves - that makes sense to me. That is the depth I am going for. However, Armstrong used PM and then converted PM to FM. I don't know how he made PM and I don't know how he converted one to the other. I have seen tons of books explaining quantum mechanics for "dummies" why can't I get some help with this? I am not looking for a complete description of everything - just a little better understanding (just as I feel like I could give your hypothetical biologist a better understanding). I will keep on searching and trying and bothering people.

I really did appreciate the help of the person who originally responded to my post. I have found this website to be full of people who are incredibly knowledgeable and generous with their time.

Anyway, at least I learned a new term "boffin" which is cool

Kathy

 

[sophiecentaur]

However, Armstrong used PM and then converted PM to FM.

There is very little to choose between PM and FM. The demodulated signal may look pretty much identical, actually. The relationship between Phase and Frequency is
Phase is the time integral of frequency.
or
Frequency is rate of change of phase
I looked for some available information about phase modulation with thermionic valves (all that was available in early days). I encountered you problem that all the hits seemed to be Patents. Here's one.
With this sort of circuit, you can produce a varying phase (proportional to input voltage signal) of up to 90°, which isn't a lot ('low deviation')
If the modulator produces a phase that's linearly related to the input audio signal and you want to produce FM, you just need to Integrate (dt) the input signal. This can be achieved with a simple RC circuit. If you want to produce higher deviation FM, (with a deviating phase more than 90°) it is possibly to take the carrier with low deviation and use a Frequency Multiplier circuit (Generate a high harmonic with a non linearity and a band pass filter) A third harmonic will have three times the frequency deviation of the fundamental. High deviation FM gives a significant noise advantage over straightforward AM.
By the time you are using FM, you will need a cleverer and more 'precise' demodulation circuit. This would require a filter, tailored to work at one particular frequency so that implies the need for a superhet with an IF frequency that's the same for all received signal channel frequencies. The superhet idea had so many advantages and would allow a more sensitive receiver with higher gain at a 'manageable' internal frequency.
Using 'mixers' and local oscillators is no big deal and has been the bread and butter of radio frequency circuitry.

If you want texts along the lines of 'Radio for Dummies' then there are loads of web pages. Just get used to searching through a lot of dross before you get what you need at your level. Radio Ham sources can be useful but there can be a lot of jargon to wade through.

 

[tech99]

All I am looking for is a basic explanation of how he made the frequency change. For example, in another thread I saw that they use variable capacitors to change the capacitance of a LC circuit to make FM waves - that makes sense to me. That is the depth I am going for. However, Armstrong used PM and then converted PM to FM. I don't know how he made PM and I don't know how he converted one to the other. I have seen tons of books explaining quantum mechanics for "dummies" why can't I get some help with this? I am not looking for a complete description of everything - just a little better understanding (just as I feel like I could give your hypothetical biologist a better understanding). I will keep on searching and trying and bothering people.

I really did appreciate the help of the person who originally responded to my post. I have found this website to be full of people who are incredibly knowledgeable and generous with their time.

Anyway, at least I learned a new term "boffin" which is cool

Kathy

Hi Kathy
I will try a simple description for you.
I am sure you will realize that phase and frequency modulation are very closely related, and believe me it is trivial to convert one to the other, but AM is rather different.
With AM, we take a carrier wave, say 1000kHz, and we take the audio, say 1kHz. Now we use a circuit which causes the amplitude of the carrier to go up and down in proportion to the instantaneous voltage of the audio. This creates a beat pattern, or modulation envelope, which you will find in many books, Wiki etc.
If we now look at the frequency spectrum of the signal we have created, we find it seems to comprise three waves, of 999, 1000 and 1001 kHz. The centre one is our original carrier, and amazingly it actually remains constant, but two additional side frequencies have popped up. If we measure the amplitude of what we have created using an oscilloscope, it is of course going up and down. But what is happening is that the three waves continuously vary in relative phase, so that at the peak they are all in phase, and at the trough the two side frequencies are in phase with each other but oppose the carrier. So we can imagine a phasor diagram with a fixed arrow illustrating the carrier and two counter rotating arrows representing the side frequencies. These alternately add and subtract.
With PM, we use a circuit which causes the phase of the carrier to be moved in proportion to the instantaneous voltage of the audio. This does not cause a change in amplitude of the carrier, and if we look at the spectrum it also has three frequencies. But the difference is that the phase of the carrier is such that when the side frequencies coincide in phase, they are at 90 degrees phase to the carrier. So the amplitude does not vary.
To convert from AM to PM, Armstrong took an AM signal and deleted its carrier, leaving two side frequencies. Then he added a new carrier which was shifted in phase by 90 degrees. This synthesised a PM signal, having all the correct properties. The amplitude did not go up and down with the audio and it had two side frequencies.
Now to explain about FM and PM. With FM, the frequency shift of the carrier is proportional to the instantaneous amplitude of the audio. Of course, changing the carrier frequency also implies that the phase is also being altered. And with PM we find that the frequency shift of the carrier is proportional to both the amplitude and frequency of the audio. So if I listen to a PM signal with an FM receiver, the sounds that are higher pitched are louder than those which are lower pitched. This can be corrected with a simple tone-correction circuit, which actually proves an audio slope which is described as 6dB/octave. In other words, the audio gain is reduced in proportion to the frequency of the sound. By using simple tone correction circuits, PM and FM can easily be interchanged.
[Note for experienced readers: For simplicity, I have used the term "carrier" most liberally, but not wishing to upset sensibilities! I have also not mentioned the additional side frequencies which exist with angle modulated systems, because in Armstrong's modulator, the deviation ratio was small, so they were negligible. I have also not mentioned that with angle modulation, the carrier is changed in amplitude, whereas perversely, with AM it is not].

 

[mrad]

This thread is a tad old, but I would add that the principle of balanced FM synthesis can be shown concisely in mathematical terms familiar to a physicist without using radio jargon. If an arbitrary audio signal is a function of time S(t), then the voltage of a PM signal with radio frequency f is V.pm=cos(2*pi*f*t+S(t)) while FM is V.fm=cos(2*pi*integral(f+S(t))dt). AM is V.am=S(t)*cos(2*pi*f*t). Integration of a signal is easy; just let charge accumulate on a capacitor. Adding signals is basically 2 resistors in parallel so the current flow adds. With a bunch of trig identities, S(t) << pi means approximately V.pm has a similar form to V.am, but some terms are shifted by pi/2 (ie, sin instead of cos). So by integrating the carrier to turn sin to cos (shifting by pi/2 that is) and adding together various voltages so that terms cancel out (a "balanced" modulator) you can get V.pm from V.am. To get V.fm you just have to integrate the input signal before phase modulating. I glossed over details like amplitude and frequency deviation constants, the latter of which scales up when frequency is increased by harmonic-based frequency doublers/triplers or by superheterodyne mixing. (Modern radios don't use this technique, but Armstrong didn't have variable-capacitance semiconductors in the 1930's.)

I can't find much on the receiver, but superhets were well understood by this time (especially by Armstrong, who invented it). Foster-Seeley discriminators weren't invented yet, and decent phase-locked-loops needed the same not-invented-yet variable-capacitance semiconductors that direct PM synthesis requires, so I have to assume the front end was a superhet much like contemporary AM receivers, but then a limiting circuit to make the amplitude uniform and get rid of any AM, followed by a slope detector to put AM back in as the frequency moves in and out of a circuit's resonance, and then an AM detector.

The actual apparatus Armstrong used now sits in a museum in the Rochester NY area. http://www.antiquewireless.org/uploads/1/6/1/2/16129770/3549218_orig.jpg

 

[sophiecentaur]

So by integrating the carrier to turn sin to cos (shifting by pi/2 that is) and adding together various voltages so that terms cancel out (a "balanced" modulator) you can get V.pm from V.am.

I have doubts about the possibility of wide deviation FM being produced with this method. It's fine for narrow band comms systems but how would that 'phasor addition' work for phase deviations of several π?

 

[tech99]

I have doubts about the possibility of wide deviation FM being produced with this method. It's fine for narrow band comms systems but how would that 'phasor addition' work for phase deviations of several π?

The

I have doubts about the possibility of wide deviation FM being produced with this method. It's fine for narrow band comms systems but how would that 'phasor addition' work for phase deviations of several π?

The Armstrong method generated FM at a low frequency and then multiplied it up many times to obtain the required wide deviation. The modulating audio was integrated with a CR network to obtain FM rather than PM. The 90 degree phase shift was obtained by balancing out the carrier and then re-injecting it in quadrature.
At the RX, it looks as if a superhet was used, with a limiter and then a discriminator using two tuned circuits, one above and one below the carrier frequency.

 

[sophiecentaur]

FM at a low frequency and then multiplied it up many times

Yes of course. I had quite forgotten that trick.