Placing the rdp-1 into the Input mode permits selection of input source, output resolution/dither, and whether the DSP operates on both channels (Stereo) or on each channel independently (Dual-Mono). Since the output of the DSP is 32 bits wide, the rdp-1 will downsample the output to a defined bit-width of 16, 20, or 24 bits. Selection of output resolution and dither will determine whether the full capabilities of the DAC will be realized. Dither, a calculated (not random) noise signal based on a statistical algorithm, is added during the downsampling from 24 to 20 or 16 bits to avoid the effects of truncation, the visual equivalent of which is the pixelation seen on low-resolution digital images. The rdp-1 offers 24-bit, 20-bit (truncated, no dither), 20-bit (dithered), 16-bit (truncated, no dither) and 16-bit (dithered) outputs. Feed your DAC more bits than it can chew and the result is truncation and harsh sound. Feed it fewer than it can handle and the result is a lower level of performance than you paid for.

And don't rely on the advertising fluff. A 20-bit DAC chip is too often mated with a 16-bit S/PDIF receiver (like the Yamaha YM3623) or a 16-bit-input oversampling filter (like the NPC SM5813), and the package is still touted as a 20-bit DAC! Beware: Such devices will lop off anything over the smallest bit limit inside, and actually sound worse than they would with a correctly formed 16-bit signal. The DACs I use are capable of processing 20-bit signals, and they sounded best to me with the 20-bit (dithered) output. The only exception was that the 20-bit (truncated, no dither) form was needed to pass HDCD signals for decoding. Indeed, the rdp-1 passes decodable HDCD in any of its nondithered modes, as indicated by the LEDs on the DACs and by my ears.

Selection of Stereo or Dual-Mono depends on the kind of processing that you intend to perform. Adjustments to low-frequency balance in loudspeakers are often best made in Dual-Mono mode, as such responses are greatly influenced by the unique position of each speaker in the room. On the other hand, adjustments to frequency response to correct anomalies or imperfections in sources are usually best made in Stereo mode, as true stereo recordings carry something of each voice and instrument in both channels. Changing one channel in a way different from the other destroys harmonic and spatial representation. With mono and pan-potted recordings, Dual-Mono mode may be of use.

Transparent (?) Tone Control
The Volume mode is an absolute delight: it covers a range from -95db to +12dB, and, in the critical ±12dB span, in exquisitely fine increments of 0.2dB. Increment size increases with decreasing gain, which makes large changes easy where fine control is rarely needed. Aided by my trusty multimeter and a test CD, I could match levels from different sources, eliminating a pesky source of listener bias.

More important, this volume control seems entirely unburdened by the subjective effects of bit truncation, at least in the usable range around unity gain. This is probably due to the high bit-width of the DSP. Raising or lowering the gain by fixed increments (with equivalent analog-domain gain compensation) was an entirely transparent operation to my ears. Even large amounts of attenuation (-20dB to -40dB) were difficult to fault, but the varying noise contributions of the source and analog components made the audible evaluation less than critical. I'm happy to say that, even though the DSP chip is always in the circuit (even in Bypass), inserting the rdp-1 into the system did not compromise the sound in any discernible way.

Actually, I think the system, connected via S/PDIF, sounded better with the rdp-1. At first this bothered me, but it soon began to make sense. I had inserted the rdp-1 into the system between the DDS•Pro transport and the Uther DAC in place of an intervening anti-jitter box (DTI•Pro32 or DragonPro). Bypassing the rdp-1 in that arrangement resulted in a soundstage less stable and deep, and a slight loss of instrument delineation. Zounds! That's the kind of observation often associated with anti-jitter boxes.

Adding in the DTI•Pro32, the presence (or absence) of the rdp-1 was not detectable. My feeling is that, in a low-jitter environment (with the DTI Pro32), the rdp-1 was completely transparent; and that, in a system with some jitter liability (without the DTI•Pro32), the rdp-1 had an ameliorating effect. That's like having your cake and eating it! (I still preferred the all-I2S arrangement offered by the Alchemy or Camelot components if EQ was not needed, by the way.)