Copycode & the Future of DAT Beware of DAT, says Peter W. Mitchell

Beware of DAT, says Peter W. Mitchell

Beware of DAT. If, despite the limited availability and high cost of digital cassette machines (and their companion tapes), you still feel irresistibly tempted to be the first on your street to invest in the new medium, buy with caution. The analog-to-digital converters in some of the first-generation DAT machines have a linearity defect at moderately low recording levels.

Their performance is fine at high levels (near 0dBFS) and at very low levels (below –70dB). But from –30 to –60dB, the encoding defect in some machines produces distortion levels greater than 1%. In one Aiwa DAT that Brad Meyer and I tested, the distortion changed with level: at –40dB the signal was accompanied by a rich, buzzy spectrum of even-order harmonics, but at –45dB the distortion included many odd-order harmonics as well.

It is difficult to know how widespread this fault is, since most people test digital recorders only at very high and very low levels. (The problem has been encountered in Japan, the US, England, and West Germany, so it's not just an isolated case.) I am reminded of the first-generation CD players from Hitachi and Denon, which had a major non-linearity at about –30dB that, as far as I know, was not disclosed by any published test report (footnote 1). Reviewers naively assumed that if the player had low distortion at 0dB and was reasonably linear at –80dB, it must be okay in between. The defect was plainly audible in music, making hash out of mid-level woodwind and piano tones. Denon quickly stopped selling the machine and stayed out of the CD player business for nearly a year. (Today's Denon players, of course, are first-rate.)

The nonlinearity is only in the encoding (recording) circuits of the DAT; the playback decoders are OK, so prerecorded DAT tapes can sound quite good. Direct digital copying would bypass the encoding defect of course, but CDs can't be digitally copied onto any DAT. If you decide to invest in a DAT, borrow a low-distortion signal generator (or use the 0dB tones on a test CD), and test the individual machine that you're buying. Record a 500 or 1000Hz sinewave tone on the DAT at –40dB, and compare the playback against the original tone. If your machine has the linearity defect, the recorded signal will sound slightly buzzy.—Peter W. Mitchell

Footnote 1: This was discussed in the review Martin Colloms carried out on the Hitachi DA-1000 CD player for me in HFN/RR, March 1983. The machines suffering from this problem in the UK were sold very quickly at bargain prices. I can't remember that any of those customers complained about what I, too, found to be quite audible distortion.—John Atkinson