Vintage

Testing the CDA-94 separately is very instructive because, while it’s based on the same core SAA7220/TDA1541 4x/16-bit chipset (the R1 version, in fact) as the CD-94, this outboard solution will accept data up to 48kHz. That said, the CDA-94’s indicators show ‘48kHz’ regardless of whether it’s receiving a CD/44.1kHz or 48kHz/24-bit input, and the balanced XLR outputs remain phase inverting regardless of the position of the ‘absolute phase’ rotary. The TDA1541 will only accept data up to a 16-bit wordlength so 24-bit inputs are truncated at –96dBFs [see red trace, Graph 1] and while it will handle up to 384kHz Fs, the input receiver here does not go beyond 48kHz (or 192kHz with the 4x oversampling filter). The ‘step’ in distortion where the TDA1541’s 10-bit passive divider bites (at –60dB) is also very obvious [again, see Fig 1, below].

With these caveats understood, distortion is still significantly lower with 48kHz data through the CDA-94, falling to 0.0003-0.0005% at its maximum 2.4V balanced output (re. 20Hz-20kHz/0dBFs) even though the A-wtd S/N ratio is about 1-bit lower than achieved by the CD-94 at 100dB. Low-level linearity is also slightly further adrift – signals at –90dBFs are some 3dB lower in amplitude and those at –100dBFs are compressed by about 1-bit. The CDA-94’s AC-coupled output also has a more pronounced bass roll-off, falling to –0.3dB/20Hz and –1dB/5Hz but reaches out to –0.9dB/22kHz with 48kHz inputs. Stopband rejection, courtesy of the SAA7220 4x chip, is 41dB with CD data and 32dB with 48kHz inputs while the impulse (time) response, with its limited pre/post ringing, is the same as shown on Graph 2, p131. Finally, jitter is somewhat higher at 520psec, and dominated by data-induced sidebands [see Graph 2]. PM