I found a very basic AM transmitter circuit on the youtube video here. The schematic is below:
I'm a bit confused on the op-amp circuit on the right. It seems that the op-amp is really only taking the difference between the message and carrier signals - not multiplying them together like it should be doing.
I'm not at all convinced that this part of the circuit is a frequency mixer. Am I just crazy?
Further to my comment. You are quite correct about needing some form of multiplier. Perhaps the following addition could be made to improve the circuit. The FET acts as a voltage controlled resistor altering the gain of the op amp creating an AM signal.
Note the presence of the 100k pot - Why is it there? Does it perform some linear function in a summing amplifier? Not really - my guess is that it biases off the output against (say) the lower power supply rail (+1.5V for the TL072).
With no modulation signal, the carrier is not (or barely) present on the output of the final op-amp at all - the 100k pot ensures this. When a modulation signal is inputted the op-amp output is "lifted" out of output saturation thus allowing the "carrier" to break through and, it breaks through with an amplitude that is governed by the modulation.
Is it hi-fi? No - the output distortion on the radio is hard to tell (because it's a crappy little radio) but it will be something like AM.
Towards the end of the video it showed the waveform and that's exactly what is happening.
First, someone posting a youtube video or a web page is orthagonal to good design. In fact, if it's not a professional web page, it's more likely to be crap. Think about it. The guy that knows what he's doing and builds a quick and dirty AM modulator doesn't think anything special about it and probably won't post a web page. The guy that spent two weeks on random trial and error is much more likely to proudly post "Looky me world, I done made a modulator!!!".
Second, AM modulation isn't a pure multiply. That would be double side band. If you express AM as a multiply, you have to add a offset. It can also be thought of as adjusting the volume of a oscillator on the fly, which is not that hard to realize in analog electronics.
Now for the circuit. A TL072 is not a great choice for 1.2 MHz signals. This is most likely what this guy had in his junk box, which is more evidence this is less than a professional design. Lack of component designators is another strong clue.
However, this circuit could actually work well enough, within the limitation of the low gain of the opamp at the carrier frequency. Note that when the pot is at center, it biases the opamp at 1.6 V. Now look at the common mode range of the the TL072 and see the large headroom it needs on both ends. I think what is going on is that the modulation is performed by clipping the carrier at a different level depending on the modulation signal. The pot is adjusted so that the carrier is just about half clipped off with no input signal. That allows the input signal to cause the carrier to go all the way from 0 to full while idling at 50%.
You're probably thinking that would introduce horrible distortion. Yes it would, but note that this distortion is in the carrier, which is a fixed frequency. When looking at the frequency spectrum, distortion shows up as harmonics, which will occur at all integer multiples of the carrier frequency. So this modulation scheme introduces a lot of distortion components with the lowest being twice the carrier frequency. However, the radio that receives this will simply reject those components because they are way out of band. This modulator makes a mess at other frequencies, but at the frequency of interest it looks like it can work well enough.
Something like this would never pass FCC tests due to the large amount of out of band crap it radiates. But if you're just doing this accross a room with a small antenna, then the crap won't go very far and probably won't cause enough trouble so that someone complains.
I haven't watched the video, but this might actually work. I haven't seen this technique before, but it is a sort of pseudo-am. They are using a difference amplifier with equal gain to both signals. The two pots are used to make minor adjustments to either the audio or the carrier gain. I have to assume this is a bit touchy to get right.
The difference amplifier topology is basically:
If you set \$\frac{R_f}{R_1}=\frac{R_g}{R_2}\$ then the output is
\$V_o=\frac{R_f}{R_1}*(V_2 - V_1)\$
You're right that this isn't multiplying, but you do end up with an AC riding on the carrier that looks like this (blue is the pseudo-am output and the red is the audio input):
If you run that through a diode detector it will pass the AC portion into your low-pass filter with a variety of distortion depending on the balance of the transmitted signal and how hard the diode turns on.
I haven't spent the proper time to really look at this. However my quick analysis suggests that it might work. But it isn't AM modulation and there might be a lot of distortion as Andy Aka mentions due to non ideal conditions (which might even help introduce AM at the audio rate).
EDIT: It may not be obvious how this could work on an AM radio. The carrier biases on the receiver and the diode. No envelope is found so a typical receiver will adjust the agc gain up. This little transmitter is putting out a very strong amplified audio signal directly into the open wire (the 100pF couping cap is meaningless to this signal because the wire is open). There will definitely be some leakage directly to the AM receiver if is it close by. Most AM receivers don't have much filtration below 500Khz on the input and the audio section which is on and running hard just has to pickup some of the coupled signal. Filtration will be strongest in the IF, however depending on the architecture you could even pick a RX frequency where the image lies in 1Khz to 15Khz (or in the 2nd or 3rd filter harmonic) and get some help from the IF amps. While no one would do this on purpose, I can see how this could be made to work in the demo without actually having a normal AM input.
Edit2: Look, if this works, it's only because AM receivers are highly susceptible to noise from a variety of sources including direct feed through, which is what I'm suggesting. The carrier is biasing on the AM receiver but there is no modulation on the carrier itself so the only thing the backend picks up is noise and it's going to crank up its gain. This allows the amplified audio to feed through (I suggested a few probably paths) which won't work reliably and probably only works for this guy's setup because they are in very close proximity. If you've ever heard 60hz hum on your experimental AM radio in the lab or perhaps your car's alternator buzzing on the AM radio, it's a similar effect.
