DACs

Naim Audio DAC

ARTICLE INDEX

Design

The Naim DAC accommodates the IPod through its USB inputs, one on the front and one on the rear panel. The maximum sample rate supported by Apple is 48 kHz so who cares really? Ok, almost all-ya'all care. I don't have an IPod or an Iphone but I-sure I-will be I-assimilated someday. When connected to other sources the USB inputs will support up to a screaming 768 kHz. A plan for the future since there is no known source for that data rate today. Only an IPod or USB Memory stick can be used on these USB inputs though. The USB connectors in the DAC prevent use of a cable to connect directly to a PC (or Mac). This is to prevent the hifi system from being contaminated by an ugly PC (or Mac!) ground plane. The other inputs are of the S/PDIF variety. Four inputs are provided, any one of which has an optical (Toslink) option, two have an RCA connection, the other two are fitted with a BNC connector. Outputs are the usual Naim DIN connector or RCA. Even the presence of these RCA connections is a leap for Naim which has long defended its proprietary DIN solution. A small switch on the back of the unit ensures that only one of the output modes is active.

Internally the jitter problem is solved in a unique way. The typical solution is to simply buffer the data. That is, capture it in a small memory with the clock derived from the S/PDIF stream and clock it out with a local master clock from a trusted, steady source. This type of solution has been around a while and it does indeed 'eliminate' jitter problems in the traditional sense. Jitter in digital playback has its worst effects when the unsteady S/PDIF derived clock is used for the actual D to A conversion. In that scenario the math that makes the conversion work is thrown out the window because it presumes a steady, perfect clock. Data buffering solutions use a local clock, not the derived clock for the D/A conversion. Jitter on the local clock might still be an issue because no clock can be perfect but it's no longer the S/PDIF interface that is causing the problem. It might cause other problems if it is so bad as to actually induce errors in the latching of the data into the local buffer. Or, if connected electrically, the noise from the ground plane of the digital source will become noise in the DAC. Naim has a variation on the buffer-the-data solution, Naim so distrusts the derived clock they opted to create a range of 10 local clocks to capture the incoming stream. The amount of data in the local buffer is used to make a call on whether the incoming stream should be clocked at a faster or slower rate.

In addition there is the usual oversampling (16x) but no up sampling. There is an otherwise excellent white paper by Naim on the design of this DAC that gives the usual explanation of oversampling (create extra samples with the value of 0) while completely failing to explain adequately how this could be better than creating samples with some sort of interpolation and making no mention of up sampling (which I can only guess is the process of creating extra samples with some sort of smart interpolation). The whitepaper goes on to describe the filters, discrete analog stages, vibration isolation and separation of power supplies. Mostly I will just summarize here and say that the paper reveals that Naim has considerable technical expertise as well as long experience in digital playback. The highlights to me are the DAC chips themselves (which are Burr-Brown 1704K as in almost all of the Naim CD players including the ultra-expensive CD 555) and power supply isolation and regulation. The 1704K DACs are ancient by digital IC standards, the problem with these chips was not in their sound but in their expense as their manufacture requires a 'trim' step as the resistors used in the actual D to A conversion require laser trimming to get precise values. Returning to the issue of power, when using the DAC standalone three power supplies are utilized, the whitepaper tells us: "The transformer has three isolated windings, feeding three sets of rectifiers and reservoir capacitors: one for the DSP, one for the clock circuits and the last for the DAC chips, I2V converters and analogue filters." Adding an external power supply such as an XPS that I already happen to own adds considerable expense as well as a bigger transformer and additional isolation as the master clocks now have a dedicated supply as does the analogue output stage. The DAC chips themselves remain connected to the DAC's own power supply. Naim also employs multiple voltage regulators throughout the design with the sensible goal of reducing noise on the power supply down to the realm of the (theoretical) noise in the music signal itself.