AM radio receiver - design

I suggest that you do the same. Consider making one frequency in one side band:

Y = sin(At) * sin(wt) + cos(At) * cos(st)

This does one frequency. I can now do two frequencies in the same side band as:

Y = (sin(At) + sin(Bt))*sin(wt) + (cos(At) + cos(Bt))*cos(wt)

@@ Then each Fourier component of noise would be defined by a single variable, A, B etc. in turn. But the components of additive white gaussian noise (AWGN) each have a random amplitude _and_ a random phase. What you've described would lead to a series of vectors with a predetermined set of phases so if two such series were added the voltages of the two components at each frequency would add. This is not what happens with AWGN because the phases of components from two sources at the same frequency are not corellated.

We can build up any collection of side bands we require in this way. If we are willing to do an infinite number of terms, we can get the noise as I suggested.

@@ ... and therein lies the error.

Chris

to

d.

You are missing the point. You can do one frequency, two frequencies, three frequencies ... an infinite number of frequencies.

I am keeping it simple. To get the phases just insert another constant in each component to adjust the phase. I thought that this would be obvious but I guess not so here it is with the objection fixed.

Y =3D (sin(At+C) + sin(Bt+D))*sin(wt) + (cos(At+C) + cos(Bt+D))*cos(wt)

Do you now agree that any collection of side bands can be made as I suggested?

Once you see that point, you then need to rearrange the terms. The fn1 and fn2 were not for the two side bands as you seem to keep assuming. They are two function that together compose the total noise.

d.

utter.

Here is a spot where I clearly disagree. Using one sideband narrows the RF bandwidth while leaving the recovered bandwidth alone. The narrow RF bandwidth has a smaller "window" where direct and reflect signals can mix, which is the cause of the fading.

I have a Sony with the chip. It growls. It doesn't hold the lock very well. The AR7030, granted at over 10x the cost, does a better job.

It begs the question why he didn't use an active filter. At those frequencies, the quality would be better. In any event, the DSP boxes can bandlimit just fine. Clearly, this paper is Lankford at his worse.

e

I forget who said it but: When stupidity will serve as an explanation, you need look no further.

At communications quality sound, a micro controller can make the filter if you want a low cost DSP.

If you have signal with all these problems we aren't talking HiFi.

You could also build a radio with a very narrow IF strip and clean the signal up a lot before you have to demodulate it.

Both of you seem to be talking as the noise arriving before the modulator. It is not. It is between the tx antenna and the rx antenna. It is not just Gaussian white noise, it include many impulse sources, single frequency transients, harmonic series from many sources, and even other tx on the same (and adjacent) frequency sources. There are frequency domain models for each of these sources. There is even a tractable model for the typical aggregate spectrum for medium wave and short wave for 500 kHz to 30 MHz and slices thereof.

design:

Unless I'm being paid, I just buy COTS. There are exceptions, such as if the product is extremely expsensive or doesn't even exist.

In the case of Mr. Lankford, I could just select a tighter filter in my radio. Of course, that doesn't do much for fading.

it to

an

band.

e

No we (or at least I am) talking about noise that gets added in the channel. The problem is that you have to do math on the stuff and relate it all to the carrier. Since how the noise got created doesn't really matter, it is best to call it all "side bands". We are talking about the signal in the IF strip heading towards either a 1N914 or a sync demodulator. By time you get there, the bandwidth is quite narrow.

I am lumping it all together and doing an FT on it. I am also taking a very long time frame so that all the different clicks and pops get included so that we can assume every frequency and phase will get in there some time. On this basis, I say it is just "random".

We can take any random function and break it into two functions that add together to make it. For my explanation I suggested that it be broken into one part that gets through the sync demodulator and the other part that doesn't. This requires that it be recast into things related to the carrier of the station being listened to. The whole sub thread assumes that we have a PLL with a very narrow band width so that it has a VCO with no noise in it.

Remember that the OP said normal AM radio. We are dealing with a normal AM radio design up to the point of demodulating the signal to make the audio. We have no noise near the harmonics to worry about etc.

to

No,it doesn't. It is a conversion to make some of the math easier.

XXXX->easier ^When we

Depends on the particular definition of narrow.

The math for two sine modulation is not much harder than that for single sine modulation. Then by induction you may stipulate that modulating functions of the form f(t)=Summation(from n=1 to 20000; a[n](t) * sin(nwT=t) + b[n](t) * cos(nwt)) will analyze just the same as for two sines.

Fine. Thus we are typically discussing single superhetrodynes with

455 kHz or 250 kHz IF strip and AGC. LO may be variable LC or synth.

For the add on a standard PLL at 4 * fc of IF for 455 kHz and doing I&Q synchronous has some interesting advantages. We can easily have each sideband demodulated separately and select either one or the sum.

Mit der Dummheit kaempfen Goetter selbst vergebens. --Schiller (from "Die Jungfrau von Orleans")