HIR Issue 9: On Going Digital
By Frogman
(note: Of interest to some, this article, except for this note and some minor typographical changes has been written on a TI-92 in the text editor. I transferred it to my 'miggy with Amiga92 and a parallel cable I built. I just opened the raw variable, found that it contained ASCII and some other nasty control chars and cleaned it up into this.)
Some time ago I purchased a microcassette recorder. I've used it to record many things, including some of the DefCon 6.0 presentations (I should have gotten the CDC's BO release on tape!) and my thoughts on certain subjects. One of them is my initial thoughts on going digital. In my own words:
And so it begins. Heh, Frogman gets personal with an electromagnetic device. Who would have thought the day would come? Digital technology and all it's wonders is no match for an analog recording, except for the fact that one's friggin' permanent and the other erodes away in the dusts of time. Immortality, that's what it's all about. Overrated media-hype. The digital immortality, still f***ing analog. In my 90's world, my way of thinking, everything is simplification. What do you do to simplify analog? What is analog? High, low. The difference between is called analog. When the difference no longer matters, 'cept that there is a difference, that is digital. Both do erode in the way of time. The only question is, how fast? The fact that there is a difference is still discernable. The fact of how much, that get's fuzzy. That is the difference. Like I said, immortality is f***ing overrated. Who ever wants to be immortal has go something wrong in their head. Who wants to be the same person for centuries, nothing changing over time? Nobody else change over time? Who wants to be the same person they were six billion years ago? That's immortality. Immortality won't last you about three years and go away. Immortality is f***ing forever. People want to live forever, they will see their friends, their family, everyone they either knew, hear of, cared about, thought about, considered dating, considered f***ing, whatever. When they are gone you'll still be around six billion years from now. Watching ---------- world. ------ ----- friggin' planet. Like I said, overrated media-hype.
<end of that rant>
Well, for the time being we can ignore the immortality aspects and such. They stem from a philosophy class I was taking at the time. What I want to focus on is the deal about analog vs. digital and their self refrencing traits. As I said, digital is a subset of analog, where all that matters is the fact of a difference in states. Converting from analog to digital is not all that hard, simply identify a regular period to measure from, mark high and low points, and thats about it.
The conversion of digital to analog is tougher, because of the missing information. To start, we need a new way to get from analog to digital that describes the wave. One way to do this is to increase the period to identify not just highs and lows, but the rise and fall values in between. But digital doesn't allow for anything but two values, right? Well, we can combine those two values into patterns to represent bigger numbers, like we do in base 10, or decimal. The old standard of digital to analog conversion in the home market was 8 bit. The analog signal was abstracted into a list of numbers representing the value of the wave at a point in a given period. Well, by narrowing the time between two gaps, we can get a good approximation of the wave, even to the point that we end up going past the abilities of out 8 bit value set. In that set we are abstracting the wave in to values of 0 to 255 or from -127 to 128. The problem with this is that small sublties in the wave are lost if they are between two values available, The way to correct this is to base the conversions on a larger set, such as 16 bits. This gives us a range of 0 to 65536 or -32767 to 32768. This gives a much clearer and accurate abstraction of the wave. If memory serves, this is about on par with the average quality that the human ear identifies.
Now, with a much clearer way of going from analog to digital, going in reverse is quite simple, just convince some DAC (digital to analog converter or circuit) to smooth out a few rough edges, and we have our original waves general look back.
The problem I identified with analog and digital (before heading into the immortality rant) are clear, when you consider the proposals I have just made about conversions. The analog wave is being approximated and abstacted into a string of numbers, which can be stored and copied in an exact form, since they are working in a bounded set, ie. the range allocated by how many bits you use. The waves exact properties are lost in the conversion, but the only known ways to reproduce analog waves is by approximation.
So, having found that digital is a subset of analog, and that the subset can be used to represent values larger than the individual digits in the set, we can guess the whole "digital is better" mindset really isn't true. Digital is what we have to do. That sounds olmost ominous, but digital is the only way to maintain exact copies of our analog world, only suffering a loss of data in the initial conversion.
Now that some chunks of theory and analysis are out on the table, if you've read this far, you may be wondering where this is leading. There are many applications for this whole topic, but in this issue I want to focus on audio, specifically analog to digital and vice versa, especially in the case of data transmission, such as modems. =] This all brings me around to the coupler Axon built.
Why don't people use couplers more often, you may wonder? Well, acoustic couplers convert signal to signal, just like a hardwired modem. But they have more conversions, and therefore more room for data loss. A modem converts the digital signal from a source such as an RS-232, RS-442, PCMCIA, ISA, PCI, or what ever interface goes from the computer to the modem. It is digital all the way. From the modem down the phone line to the next modem is an analog connection, which a basic analog signal is sent that the other modem can pull the digital highs and lows out of. From there it gets sent to the other computer through the whathaveyou interface digitally. Nice, one conversion to and from analog.
But what about the acoustic coupler? We start at the modem this time, and the initial digital to analog conversion. From there it goes through the coupler, which blasts the analog signal from a speaker. That speaker approximates the analog electrical signal, and turns it into moving air. The air in turn moves the microphone of the handset the acoustic coupler is attached to. The microphone approximates the air movement into an electrical signal, and sends it merrily on its way through the phone line, as in the strait modem to modem example. What this shows is two more analog to analog approximations to the data signal in top of what is there to start with. Those approximations can be royally screwed if the speaker on your coupler sucks, the microphone in the handset sucks, or the air movement in between is interfered with, which sucks.
And what you have learned today, I hope is that approximation is necessary, to a degree. Too much approximation ruins the abstraction of the analog data, and ruins this whole point of using digital tools to reduce data loss.
