Bandwidth of AM versus FM

[Simon]

I have just read in my course that AM radio has a lover bandwidth than FM. I can see that we choose to use more bandwidth in FM because we can but is AM intrinsically lower bandwidth?

[tggzzz]

Old school broadcast FM audio does not have a well-defined bandwidth. Having said that, yes AM does have a well-defined and narrower bandwidth.

Modern frequency modulation schemes can have a very well defined bandwith, and it can be as wide as you like (think spread spectrum). Theie is, of course a limit to how narrow the bandwidth can be, as given by one of the most important equations of the 20th century: Shannon's Law.

[emece67]

.

[Simon]

What I am saying is that in principle, ignoring standards are there theoretical limits on the bandwidths? I think the author of the course is struggling to explain the difference between theoretical characteristics and real world practical application.

[emece67]

.

[Simon]

So FM requires more bandwidth for the same transmitted data frequency?

[emece67]

.

[gf]

So FM requires more bandwidth for the same transmitted data frequency?

Basically yes, but I would rather say "... for carrying/transmitting the same baseband signal".

I 'm lost on the term "data frequency" in this context. How exactly do you define it?

[Simon]

I 'm lost on the term "data frequency" in this context. How exactly do you define it?

Highest frequency that can be faithfully transmitted.

[tggzzz]

I 'm lost on the term "data frequency" in this context. How exactly do you define it?

Highest frequency that can be faithfully transmitted.

Define "faithfully". Refer to Shannon's equation - or even the concepts it contains.

Back to your original question. Suppose you have an (audio) modulating signal with an amplitude of 1V. You could have a modulator with a 1MHz/V or 1MHz/mV sensitivity. Clearly they are both FM but have very different bandwidths.

Why is the bandwidth ill-defined? Because, from 45yo memories, the spectrum is defined by Bessel functions, and you can choose cutoff values.

[T3sl4co1l]

Also, approaching it from the other side -- AM simply has lower bandwidth because that's what it was defined to be, way back in the day.  And commercial FM has an intentionally wide bandwidth (200kHz channels), I guess because it performed better with the discriminators of the time?

Traditional voice channels are a few kHz -- same as telephone (500-3000Hz), and can be exactly that when modulated SSB.  Presumably, broadcast wanted higher quality than this, so they established it at a higher bandwidth.

There is narrow band FM (typical in VHF and higher amateur bands, isn't it?), which has the expected doubling (plus a little) of bandwidth.

Tim

[vk6zgo]

Also, approaching it from the other side -- AM simply has lower bandwidth because that's what it was defined to be, way back in the day.  And commercial FM has an intentionally wide bandwidth (200kHz channels), I guess because it performed better with the discriminators of the time?

Traditional voice channels are a few kHz -- same as telephone (500-3000Hz), and can be exactly that when modulated SSB.  Presumably, broadcast wanted higher quality than this, so they established it at a higher bandwidth.

There is narrow band FM (typical in VHF and higher amateur bands, isn't it?), which has the expected doubling (plus a little) of bandwidth.

Tim

Wider deviations perform better with any currently used type of FM detector.
For complex historical reasons, AM Broadcasting in Europe has very low audio bandwidths, compared to the USA &  Australia.

In both of those countries, the programme audio bandwidth was 10kHz for many years, but was in later years, reduced to 9KHz.
These audio bandwidths translate to transmitted RF bandwidths of 20kHz & 18kHz, respectively.

It would have been possible to produce MF "Hi Fi" transmissions with audio bandwidth out to 15kHz, but this becomes messy, as the RF bandwidth becomes 30 KHz, which is a substantial percentage of the carrier frequency at MF, necessitating heroic measures to make it work, including wide receiver bandwidth, with the consequent problem of intererence from other stations.

The twin advantages of noise rejection & the "capture effect" made FM very attractive, but the wider bandwidth required to take full advantage of this mode were not available at MF.

Thus, FM Broadcasting was moved to VHF, originally around 49MHz, then moving to the current allocation.
Plenty of available RF bandwidth lent itself to extending the programme audio bandwidth out to 15kHz, & increasing the deviation far above the low figures used with NBFM comms equipment, (to 75kHz).

This improves the noise rejection substantially.

[CaptDon]

Remember also that in the U.S.A. the commercial FM radio broadcast band deviation is +/- 75khz and yet we can also incorporate subcarriers with about 5khz deviation such as the 67khz carrier we used for remote control readback, but often used by stations for 'Reading for the Blind' and other such philanthropic uses. We also had a narrowband subcarrier at 91khz. Figure how that fits into a 75khz deviation scheme??? If I recall, all of our subcarriers combined could only total 8% of the total signal deviation so we didn't have all of them active at the same time. You don't need 75khz deviation to carry a 15khz audio signal, it was chosen as a cost/signal quality trade-off using available foster seeley demodulation schemes popular at the time.

[radiolistener]

What I am saying is that in principle, ignoring standards are there theoretical limits on the bandwidths?

On lower frequency you're limited to lower bandwidth.
For example, if the carrier is 10 kHz, the maximum available bandwidth is 20 kHz (0...10 kHz + 10...20 kHz).

But if you use 20 kHz bandwidth with 10 kHz carrier, your signal will exclusively use all spectrum from 0 Hz to 20 kHz and another transmitter cannot work within this frequency range simultaneously. This is why low frequency bands use low bandwidth standard - to allow more transmitters to work simultaneously.

For example, on LW (30 kHz to 300 kHz) and MW (300 kHz to 3000 kHz) band you can find AM stations bandwidth usually limited to 8-12 kHz.
On SW (3 MHz to 30 MHz) band you can find stations which works with 16-20 kHz bandwidth.
On VHF (30 MHz to 300 MHz) band you can find FM stations which works with 192 kHz bandwidth.
On UHF (300 MHz to 3000 MHz) band and at the end of VHF band you can find TV stations which works with 7 MHz bandwidth.

In short - more high frequency carrier has more bandwidth capacity, so it can use transmitters with more wide bandwidth standard.


Dependency between signal bandwidth and occupied spectrum bandwidth is another story, this is depends on modulation type.
Both AM and FM use about the same occupied spectrum bandwidth for the same signal bandwidth.

FM modulation has better rejection for static interference. But AM modulation is better for working with reflections, fast moving transmitter/receiver (due to Doppler shift) and long distances. This is why fast moving aircraft use AM modulation for communications.

[vk6zgo]

Remember also that in the U.S.A. the commercial FM radio broadcast band deviation is +/- 75khz and yet we can also incorporate subcarriers with about 5khz deviation such as the 67khz carrier we used for remote control readback, but often used by stations for 'Reading for the Blind' and other such philanthropic uses. We also had a narrowband subcarrier at 91khz. Figure how that fits into a 75khz deviation scheme??? If I recall, all of our subcarriers combined could only total 8% of the total signal deviation so we didn't have all of them active at the same time. You don't need 75khz deviation to carry a 15khz audio signal, it was chosen as a cost/signal quality trade-off using available foster seeley demodulation schemes popular at the time.

Indeed, PAL TV gets away quite well with 50KHz deviation, but you do start to have problems as you reduce deviation further.
Ratio detectors are quite a bit better than Foster Seeley in many ways, but don't do much better with decreased deviation, quadrature detectors are a bit better.
Even PLLs start to have problems at reduced deviations.

Yes, I am familiar with SCAs,(which are limited baseband bandwidth), but didn't want to go down that rabbit hole.

Stereo FM, even without any other such things, obviously is more than 15KHz baseband bandwidth, due to the 19kHz subcarrier and the 38kHz DSB modulated L-R subcarrier.

As previously noted by other posters, simply doubling the modulating frequency bandwidth by two doesn't "hack it", & Stereo FM stations commonly have occupied bandwidths of around 250kHz.
Having said that, which is my recollection from looking at the signal with a Spectrum Analyser, the specs are for quite a bit less (at a specified number of dB below reference). 

This stuff is all "off the top of my head", as I haven't been involved in this stuff for twenty-odd years, so I may well miss some pertinent points.

One reason the various subcarriers can be "shoehorned" in, is that the energy distribution in the baseband signal is quite low at high modulation frequencies, so there are effectively, spectrum "gaps", which minimise interference between the various signals.

All the stuff in blue is my added errata!

[Terry Bites]

Ideal AM has a finite bandwidth. FM is in theory (Bessel functions and such) of an infinite bandwidth.The received quality often depends on the pre-modulation audio processing. eg FM uses pre emphasis to overcome the limits of (older) FM demodulation schemes. Both AM, FM and DAB audio sources will be compressed and filtered before broadcast. The UK "World Service", for instance  sounds much the same on DAB and webcast as it does on a quiet AM channel. Ultimately the quality is determined buy many factors of which the modulation scheme is only one. SNR and reliability of service within a channel are generally the main concern for broadcasters not necessarily best fidelity.

Most AM receivers are a bit crap these days. RF pollution from (compliant hah!) power supplies, Wi-Fi and loads of wireless tech junk are destroying AM reception anyhow. DAB multiplexes are getting skinnier so FM is often the best listening choice. In the end though as long as you can make out the words and the music rocks- who cares!

[Ed.Kloonk]

AM Stereo was a thing here in the early 90's. The DJ's used to get on TV and show some of the CD's they were going to play the next day.

Cos you know, CD's sound better.

[TimFox]

"Regular AM", two sidebands plus carrier, with a band-limited baseband (audio) has a well-defined bandwidth of twice the baseband.  American AM broadcast regulations limit the baseband to 5 kHz.
"Regular FM" bandwidth is theoretically infinite (as mentioned above), but the sidebands fall off rapidly past a value that depends on both the deviation (change in frequency determined by maximum baseband voltage level, +/- 75 kHz for American FM broadcast at maximum peak voltage) and the baseband (15 kHz for mono FM in US, but wider when stereo and other signals are included in the modulating signal).
A classic reference to FM theory is A. Hund "Frequency Modulation", McGraw-Hill, 1942, chapter 1, with the infamous Bessel functions starting on page 17.

[T3sl4co1l]

Incidentally, what happens if you bandpass an FM signal?  Well, you get amplitude modulation of course.  Which we can remove quite easily by distorting the piss out of the signal -- so, it really doesn't matter much that this happens.  Receivers already need this (clipping, limiting) to deal with multipath and fading.

It is still of relevance to transmitters, which want to have as high (and stable) amplitude as possible, to avoid losses -- a class C amplifier might be used to deliver the bulk of the signal at high efficiency, with a minor amount of "plate modulation" to correct for the amplitude variation*.  (It may be better to generate the signal correctly in the first place, than to generate the ideal thing at higher efficiency and filter it down -- the filters would be quite bulky, -1dB at 50kW is measured in dollars of electric bill!)

*I believe there's a term for this type of amplifier architecture, alas it's escaping me at the moment.

To put it another way: generating QAM (rectangular coordinates) as phase * amplitude (polar coordinates).  In this case, the amplitude range is small, and relatively slow varying (100s kHz), so not much power needs to be spent on the amplifier.

Tim

[vk6zgo]

Incidentally, what happens if you bandpass an FM signal?  Well, you get amplitude modulation of course.  Which we can remove quite easily by distorting the piss out of the signal -- so, it really doesn't matter much that this happens.  Receivers already need this (clipping, limiting) to deal with multipath and fading.

It is still of relevance to transmitters, which want to have as high (and stable) amplitude as possible, to avoid losses -- a class C amplifier might be used to deliver the bulk of the signal at high efficiency, with a minor amount of "plate modulation" to correct for the amplitude variation*.  (It may be better to generate the signal correctly in the first place, than to generate the ideal thing at higher efficiency and filter it down -- the filters would be quite bulky, -1dB at 50kW is measured in dollars of electric bill!)

*I believe there's a term for this type of amplifier architecture, alas it's escaping me at the moment.

To put it another way: generating QAM (rectangular coordinates) as phase * amplitude (polar coordinates).  In this case, the amplitude range is small, and relatively slow varying (100s kHz), so not much power needs to be spent on the amplifier.

Tim

AM created by insufficient filter bandwidth is called "AM Noise", is a standard test during regular routine testing of FM transmitters, & is normally minimised by tuning of the PA stage, rather than more esoteric methods.
Most of the filtering with FM broadcast transmitters is done with fixed filtering at lower levels.

A very common type of architecture with VHF/UHF transmitters is "IF Modulated".

This was widely used with analog TV transmitters, where all the vital filtering was done at the Sound & Vision IF frequencies (33.4MHz & 38.9MHz  for the PAL system used in Oz).
These were then upconverted (& frequency inverted) to the VHF or UHF channel frequency required, followed by usually separarate, but sometimes combined wideband RF amplifiers.

This method allowed the manufacturer to use standard IF exciters, reducing both the number needed in stock, & the design load upon the EEs.

Interestingly, this method was not used in the NEC or NERA Stereo FM transmitters I worked with, maybe because the FM Broadcast band was narrow enough to  make it unnecessary.

[TheUnnamedNewbie]

It is still of relevance to transmitters, which want to have as high (and stable) amplitude as possible, to avoid losses -- a class C amplifier might be used to deliver the bulk of the signal at high efficiency, with a minor amount of "plate modulation" to correct for the amplitude variation*.  (It may be better to generate the signal correctly in the first place, than to generate the ideal thing at higher efficiency and filter it down -- the filters would be quite bulky, -1dB at 50kW is measured in dollars of electric bill!)

*I believe there's a term for this type of amplifier architecture, alas it's escaping me at the moment.

To put it another way: generating QAM (rectangular coordinates) as phase * amplitude (polar coordinates).  In this case, the amplitude range is small, and relatively slow varying (100s kHz), so not much power needs to be spent on the amplifier.

You might be looking for terms like "Polar transmitters"? There is also Outphasing, which is similar? Polar transmitters use the idea that an pass-band amplifier (such as the PA of an RF transmitter) is most power-efficient when it is very non-linear (Ie, no, amplitude information can be passed on since the output power is constant, only phase (ie frequency) information is passed to the ouptut. To use this you split a signal into an envelope (= amplitude) and constant-power carrier (=phase). You amplify the carrier (perhaps with a class E PA), and modulate the voltage rail of that amplifier with the amplitude information. I believe this is commonly used integrated solutions for low-speed (in modern IC terms) stuff like WiFi, GSM, LTE/4G, etc. For high data rates, you need a very, very fast regulator (as that datarate translates to high envelope speed) where the voltage regulator supplying the PA needs to actually have that bandwidth. And if you use a linear regulator there, you might as well have stuck to a more classic A/AB PA architecture in the first place, as you just moved the location of where you waste all that power.

Hybrid architectures do exist, that use say 3-4 voltage rails generated by DC/DC converters, and then your envelope just switches between the most appropriate range. You need to keep some amplitude information in your signal to do this, but it allows more efficient PAs since they can always operate in their most efficient range, and you gain overall linearity by adjusting the PA supply as needed. But I think I should shut up about this as this is not the topic of this discussion :p

[bob91343]

The Doherty system improves efficiency.  Perhaps that's what you are referring to.

With AM the carrier sits there all the time.  The modulation adds the sidebands, the power needs thereof provided by a modulator.  With FM all of the sideband power comes from the carrier so the total power is unchanged with or without modulation.  The amount of power needed for this varies and indeed is the total carrier power at certain modulations, hence the Bessel functions.

[David Hess]

There is nothing inherent to AM which limits its bandwidth to less than FM however noise sources have a larger AM than FM component so FM has a signal-to-noise advantage because of its greater rejection of noise which becomes important as the channel width increases to support a higher bandwidth.

There is narrow band FM (typical in VHF and higher amateur bands, isn't it?), which has the expected doubling (plus a little) of bandwidth.

Narrowband FM was initially defined to have a modulation index which produces the same channel width as an equivalent bandwidth AM signal so the channel widths were the same.  The same restriction applied to early narrowband digital modulation so it also used the same channel width.