Theta Digital DSPro Basic D/A processor Page 2
It should be noted that the Wadia X-32 (reviewed elsewhere in this issue) also uses two DSP chips, these the slightly more powerful AT&Ts. Although it is touted as 32x oversampling, the X-32 performs 8x oversampling in the digital domain and achieves a final 32x rate by using a staggered DAC array. A full discussion of this technique is provided in the X-32 review. I bring this up to dispel any confusion over why the DSPro Basic is 8x-oversampling and the Wadia is "32x," when both use about the same amount of computing power.
The DSP chips drive two Analog Devices AD1860N-K 18-bit DACs, one for each channel. The "K" designation refers to the premium-grade version of the AD1860, which is also used in the sonically well-respected Kinergetics KCD-40 player. The DAC is said to be glitch-free, thus requiring no deglitch circuit. A trim pot next to each DAC provides MSB adjustment at the factory after a two-day burn-in. The current-to-voltage converters (I/V) are OP42s, a high-quality op-amp IC from Precision Monolithics. Mike Moffat feels that a premium IC like the OP42 outperforms discrete circuits for I/V conversion due to its faster speed, a result of monolithic construction. A second op-amp IC, an OP27, drives the output buffer. Both these ICs have superior performance characteristics over most other monolithic devices. The output section consists of an op-amp per channel, in metal-can packages rather than the typical 8-pin DIP (Dual In-line Package). Six voltage regulators near the DACs, again in metal cans, support the DACs, I/V converters, and output buffers.
Overall, the DSPro Basic's design is a departure from most other digital processors: It uses Analog Devices DACs instead of the ubiquitous Burr-Brown, metal-can regulators and output buffers instead of TO-220s and 8-pin DIPs, digital-domain de-emphasis, and very few signal-path capacitors. The Basic was definitely not designed with a "follow the herd" mentality.
The first sample I auditioned worked fine with the 44.1kHz signal from the Esoteric P2 transport. However, the DSPro Basic had problems with the 48kHz output of the JVC DAT machine. Although music could be heard, it was superimposed on a metallic sound that increased in level with the music. As soon as I heard this, I remembered where I had heard this sound before: track 9 of the Sony Classical sampler CD. The noise is probably due to a problem synchronizing to the incoming data. Evidently, the same phenomenon that caused the DSPro Basic to make this sound during playback is the same as what occurred in the recording/mastering chain on that track on the Sony CD. In addition, after this happened, the Basic put out a loud thump when the input selector switch was toggled. Before the noise problem, no thump had been evident. Temporarily disconnecting AC power so the computer could reset corrected the thump problem, but the noise remained during 48kHz decoding.
I was dismayed to discover that the second sample also exhibited the same noise when decoding a 48kHz signal from the DAT machine. The second sample was sent because it incorporated new updates, not because of the decoding problem. I'm sure the DAT machine's data output is good, as it was successfully decoded by the Proceed PDP, Wadia 2000, Wadia X-32, and the Stax DAC-X1t. This is not the first time a digital decoder under review has had problems with a 48kHz signal. In my review of the British Fidelity Digilog converter (Vol.12 No.10), there was a ticking sound from the loudspeakers when the converter tried to playback a 48kHz data stream. I tweaked a trim pot inside the unit that adjusted the edge detection threshold and cured the problem.
Mike Moffat informed me that the problem is most likely that decoding 48kHz requires more software instructions for the DSP chips. Since the DSPro Basic uses the same ROM as the DSPre, which apparently has no problem with DAT machines, he was surprised by this development. Interestingly, I'm the first to report this problem (footnote 2).
These experiences point to a big potential problem when people start buying DAT machines (assuming they do) and try decoding the signal with their outboard converters. Since they have never tried the 48kHz input, they may not discover that it doesn't work until years later, after the warranty has expired. I therefore suggest that any purchaser of a digital decoder ask their dealer to try their unit in the store with a DAT machine. Make sure that the tape is not the prerecorded variety (recorded at 44.1kHz), but one made on a DAT machine at 48kHz.
Moving on to the listening impressions, I was not prepared for the level of performance offered by the DSPro Basic, especially at the "budget" $2000 price point. This is one terrific processor. The DSPro Basic's overall presentation was clearly more in the league of the megabuck units, rather than sounding like merely an improved version of an inexpensive decoder.
Specifically, the DSPro Basic had a textural liquidity and soundstage transparency rivaling any processor I've heard except the $12,000 Stax DAC-X1t. The sense of soundstage depth, coupled with the ability to resolve instruments in the front-to-rear perspective, was refreshing to hear in digital playback, much less from an affordable decoder. I had the distinct impression of instruments being behind one another, with hall reflections behind the instruments. There was a real sense of the three-dimensionality so often missing from digital. Listen to the percussion instruments on "Treme Terra" from Three-Way Mirror (Reference Recordings RR24CD). They are behind the other instruments, giving the impression of a soundstage miles deep. Not only was the soundstage deep, but wide, open, and airy as well. This remarkable recording, with its natural ambience and sense of space, reveals just how wide and deep a processor can throw a soundstage. Through the DSPro Basic and with my eyes closed, it was easy to be transported to the California Civic Auditorium where it was recorded.
Footnote 2: As described in "Manufacturers' Comments," this problem was due to faulty DSP software. Replacing the ROMs with new ones allowed the DSPro to decode 48kHz-sampled DAT data.