AM radio receiver block

24

Comments

  • ANDREW221231
    ANDREW221231 Member Posts: 295 Advisor
    edited September 2022

    oh yeah, so.. forgot that i made some progress on this account. basically something isn't working with the oversampling, causing sine waves from other 'stations' to be where they shouldn't, so i went all FCC on the frequencies and bandwidths and got a pretty good result!


    made the bands a bit wider at the expense of cool interference too, cause the sine tones of other channels seem to cr@p stuff up (annoying steady state tone)



  • ANDREW221231
    ANDREW221231 Member Posts: 295 Advisor

    ok, this is getting kinda AM radio like (with new embedded audio filez also)


    also @colB , i don't quite understand how you arrived at this filter stage... it seems like a fixed bandpass at around 9k hz would be incorrect, since some of the 'stations' are much higher in frequency... but the filter is necessary for the ensemble to work correctly ... 🤔


  • colB
    colB Member Posts: 761 Guru
    edited September 2022


     i don't quite understand how you arrived at this filter stage...

    IIRC, that isolates the side band that the audio is encoded on?

  • KoaN
    KoaN Member Posts: 103 Advisor
    edited September 2022

    So cool! It really feels like going through stations on a very old vintage radio,back in time!

    The only thing that may sound a bit different is this filter opening up from one station to another,which sounds a bit resonant as well? It's probably part of the whole thing though,maybe with less resonance would be less noticable.

  • colB
    colB Member Posts: 761 Guru

    That last version doesn't sound right to me at all. More like some filtering and distortion through some sort of phaser. Something is not right.

    The feel I was going for with the original requires all those whistling sounds in the background, those now seem muted or gone. but they should be there. Some AM radios don't have them so obviously, but that's because they are masked by extra circuitry. They were trying to remove all artefacts of the transmission technology. But for this kind of nostalgia trip, those are necessary IMO! That was literally the whole point of the project for me - the added noise and various types of distortion are just icing :)

  • KoaN
    KoaN Member Posts: 103 Advisor
    edited September 2022

    Hmm i guess what i loved is the different types of distortion,interferences possible...that sounded vintage to me when dialed in correctly.But yah the phasing,filtered sound that opens up when scanning doesn't sound right.

  • colB
    colB Member Posts: 761 Guru
    edited September 2022

    The distortion is cool - that's a post processing effect rather than an integral part of the model.

    For the rest, the demodulation seems to have been removed (maybe I'm missing something?), what's left is bandpass filtering and a rectifier.

    EDIT: there is no demodulation because instead envelope following is used... time for more research :)

  • colB
    colB Member Posts: 761 Guru

    Here's my first attempt at using an envelope follower for demodulation. It sound completely different from a ring modulator type of thing. Still authentic though. I think the problem with Andrews implementation was that the envelope follower needs to be in the full frequency process, otherwise it doesn't work properly. Not completely sure.

    I guess there could be more developments in terms of making the follower more 'authentic' with diode models etc. who knows :)

    I had something going that sounded great for noise and interference, because I accidentally had two different processes in series. Was 'wrong' but also 'interesting' :)

    Anyway here's what I'm talking about. No audio included, so unzip in the same folder as the previous version. There's a button to switch between sine demodulation and envelope follower. Watch your levels!


  • KoaN
    KoaN Member Posts: 103 Advisor

    All interesting!

    Retesting all the versions here i think i find the first version from Andrew,first he uploaded here,the most fun.

    Going for the most realistic is a different task too but i really like the tones it gives.

  • ANDREW221231
    ANDREW221231 Member Posts: 295 Advisor
    edited September 2022

    overall this process is finicky as that filter stage that i circled ended up not being necessary like i thought, but it had a different effect in earlier versions than in later. also the envelope detector distortion similarly demonstrated different effects... there was a lot of little changes made so its hard to say for sure where the later versions might have taken a left turn


    of course the main change was making sure the stations were stacking nice and linearly:


    with your oversampled version it seems some stations are sitting atop others


    think there's good possibility that's where the 'whistles' you were talking about come from


    I think the problem with Andrews implementation was that the envelope follower needs to be in the full frequency process, otherwise it doesn't work properly. Not completely sure.

    yeah, i think this is right. i never bothered to set that up properly, so it wasn't demodulating as much as it was just a distortion after effect. but it sounds a lot better that way in yours

    got rid of a few of the filter stages too looks like. nice!

  • colB
    colB Member Posts: 761 Guru

    overall this process is finicky as that filter stage that i circled ended up not being necessary like i thought

    hehe, wasn't only that filter. I've been looking over my older versions again, and as I pointed out earlier in this thread, there was definitely something wrong with it. Looking again, it seems that I had a half baked sine demodulation that didn't work, and I tacked an envelope follower (recification + low pass filter) demod on the end, then hacked away until it sounded plausible.

    In reality, it was never 'authentic'. The envelope follower version is maybe closer to what some real radios might do. But a realistic sine demodulator will require some more work I think?

  • ANDREW221231
    ANDREW221231 Member Posts: 295 Advisor

    it might be time to get out the heterodyne. was reading the old thread where laureano said a heterodyne gives the tuning sweep sounds you enjoy so well... it seems currently the extra sine waves populating the airspace are meeting that demand, and that's why cleaning them up you said it sounded wrong


    that's what the heterodyne can do, ring modding a station to frequency shift till one of the sidebands falls into range of a fixed bandpass filter. apparently the rectifier+filter method has some desirable properties too though, i think the real prize is in finding some way to frankenstein both together


    a frequency shifter could be helpful there, as it can provide the function of the ring mod for the heterodyne as a 'single sideband,, maybe shift the station into range as the bandpass tunes in from the other direction??

  • colB
    colB Member Posts: 761 Guru

    it might be time to get out the heterodyne

    I think it was 'superhet' super heterodyne that my original version was an attempt at modelling. That's where the extra 'local' oscillator and the static offset frequency comes from

  • ANDREW221231
    ANDREW221231 Member Posts: 295 Advisor

    for that it would need to be multiplication not addition


    one thing ive been thinking about for a while: beat frequencies... so two sine close in frequency and you get beating at the difference frequency which could be thought of as a kind of AM... what ive been wondering for a while is what is the boundary condition of this behavior? like i'm fairly certain this only applies to frequencies that are close to one another .. typically things that deal in sum and difference frequencies are non linear

    or maybe there is no boundary and you could demodulate a 400hz sine wave from the addition of 500 and 900 hz addition of sine waves? iono, i'd be very surprised if that would be the case

Back To Top