fm demodulator

[Mahyar-jafari]

hi,
I'm trying to design a fm demodulator,
I understand the concept of Foster-Seeley, pll and  quadrature demodulator , but I can't find the necessary formulas for designing for a particular frequency.

Can anyone help me?

Do you know any resource that can be helpful?

[Paul Rose]

For example, for a quadrature detector, the tank circuit would be designed to resonate at your center frequency.  This would typically be a fixed intermediate frequency ( IF ) like 10.7 MHz or 455 kHz in a superhet receiver ( i.e. the detector doesn't need re-tuning for different received frequencies ).

For formulas related to a quadrature detector, see http://www.signalpro.biz/quaddet.pdf

[Zero999]

There's an example on the CD4046 application note. The example is for 10kHz, but it isn't hard to adapt it for a higher frequency, such as 455kHz. If you need it to work at 10.7MHz, use an external VCO (e.g. an LC oscillator with a varicap diode) or the 74HC4046.
http://www.ti.com/lit/an/scha002a/scha002a.pdf

[Benta]

A 4046 at 455 kHz?
And a 74HC4046 at 10.7 MHz?

 

And that appnote is so full of misinformation that it should be taken out and shot. 

[Zero999]

A 4046 at 455 kHz?
And a 74HC4046 at 10.7 MHz?

 

The data sheets suggest they're capable of operation at such frequencies: CD4046 at least 0.5MHz, when VDD > 5V and 74HC4046 at least 11MHz, when VDD > 4.5V. Why do you believe otherwise? Are TI lying on their data sheets? That's a bold claim. Do you have any proof?
http://www.ti.com/lit/ds/symlink/cd4046b.pdf
https://www.onsemi.com/pub/Collateral/MC74HC4046A-D.PDF

And that appnote is so full of misinformation that it should be taken out and shot. 

Please give more detail.

[Benta]

Please give more detail.

I'll do that, but I am reluctant to pollute this thread. Should I open a new one?

[Zero999]

Please give more detail.

I'll do that, but I am reluctant to pollute this thread. Should I open a new one?

That's up to you. If you think the original poster will be interested, then post it here, otherwise, create a new thread.

[4CX35000]

hi,
I'm trying to design a fm demodulator,
I understand the concept of Foster-Seeley, pll and  quadrature demodulator , but I can't find the necessary formulas for designing for a particular frequency.

Can anyone help me?

Do you know any resource that can be helpful?

Many years since I did anything with Foster-Seeley, but from what I recall the circuit was LC type circuit with a centre tapped transformer, so calculations are simple LC calculations found in any reasonable book.

F-S discriminator is a dual detector on a centre tapped transformer with a LC circuit connected back to the centre tap output of the transformer and this would cause a varying phase shift to develop between the two detectors and the resultant was the audio output. But it has been a while since I studied the method. But I recall drawing loads of phasor diagrams to show what the phase angle and voltage of the various stages in the circuit under what conditions of modulation. You would be better find a reasonable textbook for this which covers FM in detail.

Other methods such as the ratio detector and quadrature detector are again LC circuits.

For a PLL detector, see the old data sheet for the NE560 PLL IC by Signetics.

[ag123]

bounce this topic as i've a similar question, i'm not really from the ee background, more of an enthusiast meddling in arduino stuff etc
i've been looking at the various specs of cd4046 and 74hc4046
and actually reviewed the appnote for cd4046 pll
http://www.ti.com/lit/an/scha002a/scha002a.pdf

however, my question and concern is that normal broadcast fm runs from 55 mhz - 115 mhz, these are pretty high frequencies
would that vco and associated pll circuits with cd4046 and 74hc4046 be able to 'work' at those frequencies?
e.g. making an fm broadcast demodulator / receiver

i've also taken a look at  NE560 PLL IC by Signetics in the link above
https://www.eevblog.com/forum/projects/fm-demodulator/?action=dlattach;attach=400895
on p200, the graph of free running oscillator frequency as a function of vco timing capacitance seemed to reach 100 mhz when the capacitance reaches 5pf or so.
would it seem to be that parasitic capacitance e.g. around 5-10pf possibly scattered around the board etc would easily pull the vco out of the fm 55 mhz - 110 mhz range?

do i need to do some kind of 'IF' by mixing and doing a lpf before i feed a cd4046 or 74hc4046?
the motivation is to use the  cd4046 and 74hc4046 as a digital fm tuner

thanks much in advance

[iMo]

however, my question and concern is that normal broadcast fm runs from 55 mhz - 115 mhz, these are pretty high frequencies
would that vco and associated pll circuits with cd4046 and 74hc4046 be able to 'work' at those frequencies?

Provided you talk about 55MHz-115MHz the answer is no.
With FM radios called superheterodynes (aka superhets) you have to get an IF (intermediate frequency), with FM usually 10.7MHz, and the FM demodulation occurs at that IF frequency.
The IF signal (could be 1, 2, 3, 4 IFs in an radio, it depends) is created in a front-end mixer where the input from antenna is mixed with an Local Oscillator (10.7MHz higher, or lower than the signal of interest) and fed via an IF filter (to maintain the selectivity) to the FM demodulator. An FM demodulator could be of various types, PLL is one of them. With those 2 chips and 10.7MHz IF you are at the very edge of the feasibility, imo.

[Benta]

First, burn that app note. It is so full of wrong information that will lead you astray. For instance, you can NOT use PFD II without an active filter, but Figs. 12, 14 and 15 suggest you can.

For your question, which is actually more like three 

Broadcast FM is 88...108 MHz and what you normally do is to use a tuner that brings the modulation frequency down to a fixed IF, mostly 10.7 MHz.

This is done by mixing the desired received signal with a local oscillator (LO) operating with an offset of 10.7 MHz. So your LO would either operate over 77.3...97.3 or 98.7...118.7 MHz.
This LO is today almost always a PLL controlled synthesizer for digital tuning. Channel spacing is 50 kHz, which is what both the 4046 and HC4046 PFD can handle. Obviously, the VCO will not work, you need an external VCO.

The demodulation at 10.7 MHz is the second PLL circuit, and neither 4046 or HC4046 will be up to this. Look for specific FM demodulator chips instead. Search for "quadrature demodulator".

[ag123]

thanks !

i'd read up more to learn about mixing and hopefully try a design using *4046 PLLs
square wave oscillators doing 10mhz are pretty easy to do, i'd guess sinusoidal oscillators shouldn't be too difficult with crystals either
but i'd guess using crystals would leave me with a fixed freq.
perhaps another way may be to play with the vco itself and possibly generate the 10mhz demodulating signals

i think i'd look at the external vco suggestion as well, but that actually partially defeats the purpose of using the 4046
i think Benta is probably right about the LO part being  77.3...97.3 or 98.7...118.7 MHz, as i understand if we mix frequencies close to those ranges, it would form beats and filtering (lpf) that gives the beats signals at least

i found some pretty good resources on the web
https://www.radio-electronics.com/info/rf-technology-design/mixers/rf-mixers-mixing-basics-tutorial.php

[iMo]

The CD4046 (or 74HC4046, or 9046) chip consists internally of several parts - 2 different phase detectors, VCO, charge pump, etc.

With LO (ie 98-118MHz) you may use the phase detector, you must add an external VCO (tunable from 98-118MHz), also you would need an external programmable divider (frequncy tuning) and crystal reference oscillator (with the divider to channel spacing freq).
The phase detector will work at the channel spacing frequency - 100kHz for example. That is pretty low frequency and it will work fine.

With FM demodulator you may try to use the phase detector and the built-in VCO (as the internal VCO may work around 10.7MHz). The phase detectors may work at 10.7MHz too. You have to try.

There are dozens of dedicated chips for both LO synthesis and FM demodulation - you have rather go that way.

[ag123]

if i understand things correctly, it seemed there are 2 parts, the RF part and the IF part.

RF needs to be mixed with LO (say 98-118MHz) that would produce IF say offset at 10.7 mhz (difference of frequencies)

then from this IF (do we still need a LPF here?) we can use a PLL to demodulate this signal say centred at 10.7 mhz
(yup it would seem that which would be at the edge of feasibility considering the *4046)

do i make sense?

but doing it this way may mean that the LO part is analog or 'manual' tuning unless i use some sort of a VCO for the LO as well

these days we're pretty 'spoilt'  e.g. TDA9809M does a 'single chip' solution

[iMo]

The LO could be fully analog if you wish (ie an NE612 - mixer plus oscillator,  with its oscillator tuned by a capacitor, or a varicap), but it must be stable such it will not walk out the channel (ie stable within 100kHz - the channel spacing differs from country to country).

You need an 10.7MHz IF filter (small 3pin ceramic stuff) as otherwise you will hear all possible crap off your speaker. The part of IF is an amplifier, which has to amplify the IF signal, such the IF signal level is suitable for the processing in the FM demodulator.

The IF's output is then demodulated by the FM demodulator. At the demodulator's output is the AF amplifier with the usual speaker. This all is a simplified version, there is other stuff still missing (in case you target a high-end hifi stereo rig)

PS: we talking here the FM broadcasting - the channel spacing is 100-200kHz. There is FM used in professional and ham radio use - there is channel spacing 25kHz, 12.5kHz, 6.25kHz, 5kHz. In such a case you must use digital synthesis for the LO as you can hardly build an analog LO which will be stable enough not to walk out the channel.

[Benta]

PS: we talking here the FM broadcasting - the channel spacing is 100-200kHz.

While I agree on most of what you write, this is wrong. It may be so in your part of the world (which is secret, it seems), but there are lots of FM broadcast stations operating at odd frequencies, eg, 93.85 MHz or 101.15 MHz.
50 kHz is the channel spacing to aim for.

[iMo]

from wikipedia

An ITU conference in Geneva, Switzerland, on December 7, 1984, resolved to discontinue the use of 50 kHz channel spacings throughout Europe.[2]

It could be I've been somehow biased with that part of ITU region 1

[PA0PBZ]

Channel spacing (or better: raster) can be 50KHz but stations are always at least 200KHz apart, because otherwise the deviation (75KHz)  would make them overlap.

[Benta]

Channel spacing (or better: raster) can be 50KHz but stations are always at least 200KHz apart, because otherwise the deviation (75KHz)  would make them overlap.

Not relevant, but suitable for derailing the thread. When I say channel spacing, it's from the viewpoint of the LO. The LO will never have a say in frequency allocation.

Concerning ITU and "throughout Europe", the world is a bit bigger than that. The "pseudo-OP" from today is in Singapore.

[PA0PBZ]

The "pseudo-OP" from today is in Singapore.

https://en.wikipedia.org/wiki/List_of_radio_stations_in_Singapore

[Benta]

The "pseudo-OP" from today is in Singapore.

https://en.wikipedia.org/wiki/List_of_radio_stations_in_Singapore

So? I smell trolling here. How do you know for which market the receiver is? Stop. Now. 

[PA0PBZ]

So? I smell trolling here. How do you know for which market the receiver is? Stop. Now. 

You started talking about Singapore when I mentioned channel spacing and now this is your reply?
Who's trolling here I wonder 

[Benta]

Please reread my reply in post #18. And be sure to read ALL the words. No more from me until we're back on track.

[PA0PBZ]

50 kHz is the channel spacing to aim for.

I just disagree with 'channel spacing' because the channel spacing is never less than 200KHz. What you call 'channel spacing' is raster or clicks or steps of the LO. Can we just agree on that without using words like 'troll'?

[NiHaoMike]

Isn't the usual solution nowadays is to digitize the IF signal and use some form of DSP to decode all the information in the signal, as in not only the main audio but also subcarrier audio and RDS? (And even digital audio in some cases.)