FM Transmitter

914

13:40:26

Definitely not perfect. It could use a bit more circuitry on the input so it can accept standard audio sources and a bit more circuitry on the output to add filtering and control the the transmission power. I would add these but that would make the circuit wider and the workspace here is already just barely wide enough.

The 1MHz source is an example audio source. In reality no audio source will give such high frequencies, but it helps with the seeing the modulation in the simulation. The 10mV input produces a frequency deviation of about 50-60kHz. I chose this amount because the maximum allowed deviation is 75kHz. If you want to see the deviation on the scope then you can increase the input to 100mV so that it deviates by approximately 500kHz around the center frequency of 90.7MHz so the scope shows frequencies in the range of 90.2MHz to 91.2MHz. Also I should mention that the potentiometer controls broadcast frequency and covers the whole FM spectrum plus a little bit extra on either end.

To make the input compatible with more normal input sources, it should have another common-collector buffer stage to increase the input impedance, a more sensible input capacitor value (e.g. 47μF), something to prevent high frequency signals from the oscillator from leaking back to the audio source, and maybe even some kind of peak limiter and/or voltage divider to keep from over modulating beyond 75kHz.

As for the output, just a couple more inductors and capacitors should do the trick for filtering out any harmonics. It could also potentially use another output stage for controlling the output signal strength, or maybe just a modification to the current one.

Note that none pf this has been constructed in real life. All this was designed strictly in the simulator and I have no clue how well any of it will work in real life, although I imagine that with the right components and maybe a little tweaking of the exact component values to account for unexpected parasitics it should work just fine IRL. (Also make sure to not feed it too much input signal or it’ll over modulate).

published 4 years ago

thebugger

4 years ago

I wouldn't recommend using this type of modulation technique in real life, as it relies on the base-collector capacitance and more specifically on the Miller effect to modulate the output's frequency. This is of course not very linear, and would result in high modulation distortion. In fact, I've tried this technique, it works, but the quality is very bad and there's a ton of spurious frequencies transmitted along with the fundamental.

thebugger

4 years ago

I would suggest adding a parallel varicap to the LC circuit of the oscillator, and biasing it at its most linear part of the characteristic from the datasheet. If you're using PLL, you can also add a second set of varicaps for the PLL control, and use the first set at a fixed bias at the most linear part for the audio modulation only

thebugger

4 years ago

By the way, the Miller effect is also the driving force here for the frequency shift, as even though the amplifier is common base, the gain is controlled again from the base (and therefore the base collector capacitance). Only difference is that the frequency response is not affected as there is no changing input impedance due to the Miller effect

jason9

4 years ago

I don’t believe the lack of linearity will pose a significant distortion problem to the audio signal. Sure, it’ll be slightly distorted, but I just can’t see it distorting it enough for the transmission to be less than perfectly clear (although the distortion may still be audible, especially if you have an undistorted version to compare to). I know that the modulation strength might even be as much as 10x higher at the 108MHz end than at the 88MHz end, but when the modulation varies the signal between 100.175MHz and 100.025MHz I really don’t see the nonlinearity having any significant effect. Perhaps what you constructed had a much stronger nonlinearity or maybe some other problem (e.g. soft clipping of the audio source). And regarding the miller effect, it doesn’t apply for common base. Given that the base signal is far, far lower in frequency than the oscillations it can be assumed to be shorted to an ideal voltage source. This means that any collector-base and base-emitter capacitance are as good as connected straight to ground, so their influence is ruled out. That leaves only the collector-emitter capacitance (in this case 100pF). The miller effect only concerns capacitances between the input and output of the amp, and common base has the emitter as the input and collector as the output so this is the only capacitance we have to worry about anyway. The miller effect also requires a negative voltage gain in order to work, but common emitter has a strictly positive gain since increasing the voltage at the emitter decreases the current through the transistor which can only increase the voltage at the collector. Therefore, it cannot be the miller effect since the gain is not negative. Perhaps it can be considered some weird form of the miller effect for positive gain, but it’s definitely not the miller effect as traditionally described.

thebugger

4 years ago

Read a bit more on the topic and try it out in real life. Just set up the oscillator stage with a BF494 or 2N2222 and don't use an antenna. It should still be picked up by a nearby radio. The distortion is extremely high, and clearly audible even with the most linear transistor and an undamped LC tank by the antenna. This is why some designers use a second set of varicaps (other than the PLL control set) to bias at a linear region and use only with the audio signal. Even in these best circumstances, there's at least 0.5% distortion on the transmitter side alone, let alone at the receiver end. The proposed oscillator by you is handy for FM bugs, as it's simple and there's no issues with a distorted voice, as long as it's understandable. For music it's absolutely horrible.

thebugger

4 years ago

As for the Miller effect, there's no such thing as negative gain. Amps either have gain or attenuation. The Miller effect is most prominent on common emitter amplifiers due to the phase shift between input and output and the subsequent decrease in gain at higher frequencies due to the imperfect input impedance. What you're describing as negative gain is actually the 180dgs phase shift between input and output. The Miller effect describes a relationship between a decreasing input impedance at high frequency in CE amps, that is dependent on the gain. We don't care about this specifically at the moment. All we care is that you control the gain through the base, thus altering the miller capacitance, the base capacitor shunts the miller capacitor to ground (and back to the inductor through the supply rail), thus adding capacitance to the tank circuit which alters the frequency.

jason9

4 years ago

Yes, 180 degrees phase shift is what I meant by negative gain. But like I described above, the miller effect can’t happen since there’s no 180 degree voltage shift, and especially can’t happen to the 5pF parasitics between the collector or emitter and the base because the base voltage is fixed. So there is no miller capacitor, just normal parasitic capacitance (by miller capacitor I assume you mean the extra capacitance form the miller effect, not the capacitor that the miller effect amplifies the capacitance of). So I know that the collector-base and base-emitter capacitances do not vary because of the miller effect since there cannot be a miller effect there. And the sim always has the parasitic capacitance set to 5pF no matter what. So, that means that the change in frequency cannot be related to any capacitance at the collector-base or base-emitter junction. That means only the 100pF between collector and emitter matter, and this is affected by the transistor in what may be called a variant of the miller effect, but it is not the miller effect as typically defined. So, to conclude, all capacitances to ground are fixed and non-changing. The only capacitor that may be changed by the miller effect or some variation of it is the 100pF between the collector and emitter.

jason9

4 years ago

And I still don’t understand the mechanism for which the audio is distorted. Even with extreme nonlinearity it’ll still be mostly linear just because the modulation range is so small. The only way I can imagine the nonlinearity happening is if the frequency change for audio modulation and for tuning the transmission frequency are separated so that the audio can be nonlinear independently of the nonlinearity in the transmission frequency adjustment, similar to how you describe with using two varicaps.

thebugger

4 years ago

Usually transistors are not made for the purpose of being a variable capacitors. In fact, this is considered an adverse effect and manufacturers strive to minimize or mitigate it. This is why the capacitance range is not linear and changes in capacity in a given range can widely differ to the capacity in another voltage range. Imagine a situation where an upper voltage swing pulls the frequency in one direction by 40kHz and the same swing in the other direction pulls the range only 10 or 20kHz. The reproduction will not be an accurate representation of the original signal, but a very amplitude-distorted version. Varicaps in contrast are made for the specific purpose of linearity in a given range. In fact even good varicaps have some non-linearity and most good transmitters use two sets of varicaps. One for the PLL control and the other for the audio modulation in the most linear range of the varicaps.

jason9

4 years ago

Well, I’m sure the variable parasitics can be effectively removed by adding a larger discrete capacitor in parallel to the parasitic one so that even if it varies between 5 and 1 pF it’ll instead vary between maybe 35 and 31pF. Of course the rest of the oscillator will have to be re-tuned but it should still work basically just the same, only minus most of the variation in collector-base and collector-emitter capacitance (and this won’t affect the VCO effect any because the sim predicts a VCO effect just fine despite not including variable parasitic capacitances).

thebugger

4 years ago

You'd still only affect the ,,active capacitance'' which is very non-linear. You just restrict the deviation this way, nothing else.

thebugger

4 years ago

By the way, the simulation works in a bit different way than reality, you're right. Changing the base current changes the Vceo, therefore the maximum collector swing. Lower swing means less current through the collector-base capacitor, thus a smaller apparent value.

jason9

4 years ago

Well, regardless of the mechanism, it does still create FM modulation and as long as we don’t have a real model to probe I don’t think we can actually say anything about the amount of distortion. So until such a model is made, I don’t think I have anything more to say at the moment.

thebugger

4 years ago

That's what I'm telling you. I've made such a receiver, don't work good. Audio sounds compressed and distorted, there's no accurate attack of the signal, as the highs and lows are distorted due to the non-linearity of the transistor capacitance. Give it a try yourself :)

thebugger

4 years ago

Transmitter*

add comment in editor