dsi@unccvax.UUCP (Dataspan Inc) (08/12/85)
> 1. What are the advantages/disadvantages of linear vs companding DACS? > 2. What, in the recording industry, is considered 'studio quality'? > In particular, I am talking about DACS again . . . I didn't know there were any 'companding' DA/AD converters beyond the 8 bit level. Companding converters are used in telephone circuits to increase the effective resolution at lower signal levels. As far as I know, the EIA-J standard digital audio is 16 bit, linear quantisation, with a Nyquist of 44.056 kHz. (As a side note, we were trying to figure out why this sampling frequency was chosen. It is 4/325 of the NTSC colour burst frequency ** nominal ** value, but this is too much to ask?). Stay with linear DAC's, they are very cheap and plentiful. In addition, you can always gut a CD player for the postaliasing filter (sin(x)/x correction) components if you choose 44.056 kHz. The Burr Brown 16 bit audio dac is now available in plastic although it is guaranteed to have no missing codes to the 14 bit level. There are no arguments for 10 or 12 bits if D/A conversion is all you plan to do. Unless and until someone shows me an A/D converter that is 1) guaranteed monotonic and 2) guaranteed to have no missing codes for 16 bits AND 3) a full-power digitizing bandwidth of 22 kHz for a sampling rate of 44 kHz (this is tough, folks) you'll find the state of the art at 14 bits. Then, there is this other nasty problem which often shows up in video systems using flash ADC's. The comparators used in A/D converters have far better resolution (hopefully) than the converter resolution. However, you can sit at a threshold and have, say, an 18 -bit small signal corrupt the least significant bit. We always digitize n bits (in video) but corrupt the input signal with random noise to break up correlated noise so that the LSB toggles at random. I think the same things are being done in the audio field, but that the least 2 significant bits in a 16 bit system get corrupted. For electronic music, suffice it to say that a 16 bit converter system, PROPER prealiasing and postaliasing filters (pleeease!) and the option to insert random noise (as in random noise, not an aesthetic judgement on music) is probably the way to go. 44.056 as a sampling frequency allows you to gut a cheap CD player for some sort of rudimentary filtering. > 4. What is the importance (in the studio) of SMPTE time coding, where > related to synthesizers. I would think that perhaps videos of Aside from slavelocking two audio tape recorders (or an atr and vtr) the SMPTE code is probably not that widespread in direct synthesizer connections. You only get an update every 1/59.94 second. Then again, I could be totally wrong about this. I know that you purists out there are going to broil me for these remarks, but perhaps there should be a national challenge to determine just how many bits really are required. I see a lot of people carping and moaning about 'glassy CD sound' or an insufficient number of bits at the 16 bit level these days. Keep your 6DJ8 'glare' arguments to yourselves... Because objective measurments provide a rational measure of progress... David Anthony Chief Development Engineer DataSpan, Inc.
jj@alice.UUCP (08/14/85)
I have to quibble about 14 bits being the SOTA in ADC's. What about Analogic MP-2735-1? guaranteed 15 bits, 50kHz. Needless to say, this is not a reccomendation and I do not represent my company. Hmmm, come to think of it, what about Analogic MP1926? It's a full 16 bits, you have to buy a separate S/H, which is available. Still not making any reccomendations, -- TEDDY BEARS MAY BECOME EXTINCT! HELP AN ENDANGERED SPECIES! "It was great when it all began, ..." (ihnp4/allegra)!alice!jj