DIY digital audio tape

[mbates14]

I got bored, and I feel like building a homebrew digital audio tape recorder.

I have a 32khz modulator/demodulator for A/D and D/a conversion. MC3418

Is there a way to take the digital output and store it on a standard audio compact cassette? (like the old commodore tape drives), or do you have to use audio tone frequency shift keying, and where is a circuit?

[dhaen]

Not without outrageous longditudinal tape speed and re-designed head, or a complex helical scan system as used in VTR's and DAT's etc.
To appreciate why, think about the recorded wavelength on tape.

Sorry...

[mbates14]

Couldnt I use an audio frequency shift keyer?

[dhaen]

Audio is not binary, it needs lots of numbers to describe it's levels. To encode these numbers frequently enough you need a high bit rate. That's a lot of information.
Have a look at the Bell Labs page on Information Theory (Shannon) here . Be sure to run the shockwave window "Learn more about entropy".

[mbates14]

lol. dude. Of course audio is not binary.

What I meant was, use audio frequency shift keying, such as maybe a 1200hz tone for a 0 and a 2400hz tone for a 1, and since the output of the ADC is serial made up of bits of 0s and 1s, then I could have each audio frequency represent a state, and then maybe output the states back from the tape into a tone decoder PLL and give me back my 0s and 1s. goes into a DAC at the same clock rate, and there is my audio.

That would work, no?

[dhaen]

When you read a little about Information Theory you will realise that your tones are far too low in frequency. Accomodating tones of sufficiently high frequency to encode the audio, and refresh at least every 22 microseconds will constitute even more information than professional formats use. A lot of money and effort has gone into developing digital audio recorders, and they've optimised things far far better than you or I could.
Please read it up. This particular theory is an underlying principle of all we do. Without it, you are unlikely to understand why things are as they are.

[macboy]

What I meant was, use audio frequency shift keying, such as maybe a 1200hz tone for a 0 and a 2400hz tone for a 1, and since the output of the ADC is serial made up of bits of 0s and 1s, then I could have each audio frequency represent a state, and then maybe output the states back from the tape into a tone decoder PLL and give me back my 0s and 1s. goes into a DAC at the same clock rate, and there is my audio.

That would work, no?

No.

Consider how many bits per second you can encode with 1200/2400 Hz tones. Let's be optimistic and say that one half of a 1200 Hz cycle will represent a single 0 bit, and one full cycle of a 2400 Hz tone (becuase that is the same length as half a 1200 Hz one) represents a 1 bit. You can achieve up to 2400 bits/second. Now consider that you need 16 bits to encode a single sample of high quality (CD quality) audio. You can get 150 samples/second. Apply nyquist's sampling theorem, and you can achive a max of 75 Hz bandwidth. And that's just one channel. Maybe enough for your subwoofer, but not enough even for voice (3000 Hz) or hi-fi audio (20000 Hz). CD audio uses a bitrate of: 44100 samples/second * 16 bits/sample * 2 channels = 1411200 bits/second.

Sony once made a device which could use a VCR as a digital audio recorder, encoding the digital data into a video stream to be recorded. This was only possible because a VCR records video with about 2 MHz of bandwidth.