Electronic Circuits and Projects Forum

You need to rectify it as well, then filter it.

Ah, yes. Rectifying first.

Actual implementation as I propose would it be hard to do?

What opamps should I try?

Originally Posted by atferrari

Ah, yes. Rectifying first.

Actual implementation as I propose would it be hard to do?

It would be fairly pointless, the whole idea of AM modulation is that the carrier frequency is considerably higher than the data, so a simple RC filter after the rectifier is all that's needed.

The output of a 455KHz IF filter when receiving an AM broadcast is not the same as a sampled signal. An active filter can be one of many configurations so you should specify one of the basics, like high pass, low pass, bandpass for example. In any case, if you operate the op-amps in their linear range then you will not recover the modulation. A non-linear process of some sort is needed.

Ron,

It seems that I failed to convey the idea of a filter. What else if not an LPF?

Pity that I cannot try that right now.

Originally Posted by atferrari

It seems that I failed to convey the idea of a filter. What else if not an LPF?

As I've told you repeatedly, you MUST rectify it before filtering, obviously a low-pass filter is what's required, but an opamp active one isn't needed.

If you look at the spectrum of a 450Khz IF signal from an AM receiver, you will see a carrier at 450Khz surrounded by sidebands extending perhaps 5KHz either side of the carrier. Those sidebands contain the information that you want. The problem is that you need to move those sidebands from 450KHz down to 0 Hz. This is essentially what AM demodulation does. Passing the signal through a low pass filter will not move the sidebands down to 0 Hz. In order to do that you need to perform a mathematical multiplication of the incoming signal with either a carrier from a local oscillator (the old BFO function found on shortwave receivers), or more simply by multiplying the incoming signal by itself. Here is the math: Multiplying a sin(2*pi*f1*t) by a cos(2*pi*f2*t) results in 1/2*(sin(2*pi*(f1+f2)*t) + sin(2*pi*(f1-f2)*t)). so, you end up with the sum (in frequency) of the two incoming signals and the difference (in frequency) of the two. Now, if f1=f2 then you end up with the signal at 2*f2 plus another copy at 0Hz. If one of these signals has sidebands, those also end up around 2*f2 and 0Hz. If you mix an incoming AM modulated carrier with itself, the sidebands retain their relationship to the carrier and so you still get a clean copy of the sidebands on either side of 0Hz. A low pass filter can then pass these sidebands that got moved down to around 0Hz (baseband). This multiplication process can only be done with a non-linear function. You can use a mixer, for example. The simplest type of mixer is a single diode, which is the classic simple AM demodulator.

Clear now. Thanks to you both.