FM radio design resources

I assume this means that the sound will come out all wrong.

Not necessarily. You can put an FM signal through a frequency multiplier, for instance, and get perfectly good sound nonetheless.

Thing is, the waveform is an amplitude thing, whereas the modulation is a phase thing.

Cheers

Phil Hobbs

Interesting! In that case, I shall proceed onwards with constructing the receiver, and deal with new problems as they arise.

Well time has gone by and I have got the PLL working! Input to a VCO is green, output from the PLL control signal is blue:

It locks from between about 10-11MHz and can be used to demodulate FM audio just fine. For the final "productionised" version I would like to remove the negative rail I am using for my loop filter amplifier. Currently the loop filter is a simple two transistor diff amp: in order for the input to go to 0V, this requires a negative voltages source which for a battery powered device is not ideal.

My question is: What is the simplest possible amplifier where the input can go to the negative rail?

From some reading in the "Design of Analog CMOS Integrated Circuits" I understand that something may be possible with a pmos diff pair, but I cannot get anything to work in simulation. @Phil Hobbs any ideas?

Nice.

Well, there are lots of RRIO CMOS op amps out there, e.g. . If you want something discrete, a PNP diff pair with a resisitive pulldown would work, and be a lot quieter than discrete small-signal PFETs, which are pretty scarce nowadays anyway.

Something like that obviously won't pull down very strongly near ground, but then all you're driving is a varactor, right?

Cheers

Phil Hobbs

Harry - there are numerous op amps that allow the input below the negative rail. I can't recall a specific number though. The rail to rail devices Phil mentions might be a good starting point for a search. These have, I believe, both p & n channel devices on the input. The ones I vaguely rember were input from below negative to within 2 volts of positive.

Hul

Harry Dudley-Bestow snipped-for-privacy@gmail.com wrote:

Thanks for the input.

Hul I am aware of rail to rail input amps :). What I would like to do for my project is make something using discrete components.

I will try out the PNP diff pair and see how it goes. I am not so sure where this "resistive pulldown is supposed to go but all will become clear, I'm sure.

Two PNPs connected emitter to emitter, with the bases connected across whatever phase detector you're using. (It'll need to be biased above ground if you want any significant output swing from the amp.)

Resistor from the emitters to the supply to provide bias current.

One collector grounded directly, the other connected to ground via a resistor that provides the output voltage to the varactor.

Some appropriate bypass capacitor across the resistor to give a generous PLL bandwidth without letting much of the RF get lost in the resistor.

Cheers

Phil Hobbs

Thanks for the clarification Phil. That's more or less what I had going in LT spice. The tip about biasing the input higher though got the circuit to work in the sense that it provides some gain, but that's not sure what I want. here's what I have now:

reference here is the circuit I'm using now that works just fine, the only downside being the requirement for a negative rail: I'm using an XOR phase detector (The double balanced mixer gave me no end of trouble) and so the output of that after filtering is 0->2.5V and experimentally I've determined a gain of about 5 is what I want. Correct me if I'm wrong here but the problem is that the output of that circuit can't be zero because of the quiescent current through the resistor on the collector of the pnp (Rc2 in my schematic). I can bias the input voltage to be higher with a simple pullup on the base of the input pnp, but I really need the output to be 0V when the input is 0V, else my varactor will have too high of a voltage on it to begin with. The output range also doesn't seem that crash hot (1.1->2.2V) either but I presume that I can fix that by fiddling with the current source and the output collector resistor.

If you're OK using an XOR gate, just use an op amp and move on.

Alternatively, gain up the 5V swing, and use that directly with an RC lead-lag filter. (For FM demodulation you don't care about reference frequency ripple, after all.)

Otherwise, putting down the mouse and doing a little bit of algebra will make things clear in a big hurry. Those sorts of questions are easy to figure out with a bit of thought.

Cheers

Phil Hobbs

It wants to verify i'm me. Just to view a photo. So screw it!

Sorry about that John, I checked the link in an incognito window so it should work without a login.

Phil the XOR gate was just an experiment to try and get to a working design and iterate from there. I have now got everything working again with a two-transistor NAND gate. This is of course the same as an XOR gate only it goes from VCC/2->VCC instead of 0->VCC. Now I don't have any problems with trying to make an amplifier where the input goes to the negative rail :D. (I couldn't get that to work anyway, even with pen and paper). This also means that a simple current mirror can be used to bias the diff amp that previously required a negative rail. The output of course no longer goes to the negative rail either, but a couple of diode connected transistors takes care of that!

Overall the design isn't particularly elegant but it does work whilst sticking to my original principles, and that's enough for me at this point. I just want to finish the damn thing.