Technical Director of Electronics
An interview with Naim Audio's Steve Sells
Steve Sells is Naim’s Technical Director of Electronics, overseeing the development of Naim’s electronics, including the already-legendary Statement amplifier, and the company's new network players, We asked him to clarify some of our questions about the ND 555 network player.
Audio Esoterica: So the ND 555 — could you possibly make it any heavier?
Steve Sells: There’s another trillion kg of copper in the ground waiting to be made into Naim brass suspension systems... so we could have! Such self-control…
AE: In a nutshell then, what are you most proud of, regarding the ND 555?
SS: For me it’s the sound performance and hearing the evangelical feedback from customers. They have been patiently waiting for the 555 to add 500 series quality streaming to their 500 series systems.
During development it was seeing the new ideas come to life from ‘sketch books dreams’ to reality, such as the master clocking, the nested faraday cage and the DR regulators directly below the DAC.
AE: The master clock arrangement applies only to network sources, rather than SPDIF/USB inputs — does that mean the ND 555’s best performance is via the network? This will include Roon-controlled playback, presumably?
SS: We do find the UPnP is typically the best sounding source; however great care is taken over S/PDIF. We did two years of research and design during the development of the original Naim DAC to optimise S/PDIF conversion. We designed a switched ‘fixed clock’ buffer system. For all intents and purposes it plays and processes the digital audio as if the DAC was the timing master and not the slave. The ND555 uses our third generation of the same principle and isolates 100% of the incoming jitter from the S/PDIF to the DAC. It’s a super Roon end-point too.
AE: Are the advantages greatest when playing high-res or equally valid with CD and below
SS: Master clocking minimises timing errors irrespective of the incoming signal. Certainly for lower than CD quality there’s bigger effects to worry about.
AE: LVDS (Low Voltage Differential Signalling) is mentioned as being used for I2S in both clock signals to the streaming card, and for audio data, is that right? Can you explain the advantages? — some might think lower voltage would mean higher relative noise.
SS: I can see how this can appear counter-intuitive. However the LVDS digital signal is relatively robust, unlike an analogue signal. By reducing the digital signal’s amplitude, the electromagnetic and PSU noise is reduced. The sensitive analogue circuits then operate in an electrically quieter environment. The LVDS digital transmission is perfectly transparent and will not drop a ‘bit’.
AE: From your white paper: “The task of the analogue output filter is to provide sufficient attenuation by this frequency to reduce the content of image frequencies in the ND 555’s output to insignificant levels.” Can you explain “image frequencies”?
SS: Another term used to describe the ‘image frequencies’ is ‘beat frequencies’. The best analogy we’ll have all heard is a twin propeller aeroplane. For example if one engine is rotating at say 1000rpm and the other at say 1060rpm there is a 60rpm difference — the same as one hertz. Here you would hear the propeller noise beating at 1Hz making a huummmMMMmmmMMMmmmMMMmmm’ sound.
Back to digital audio: before digital filters, beat frequencies were difficult to engineer out. In these very early first generation CD players playing at 20kHz beating with the sample rate of 44.1kHz would make a tone at 24.1kHz that the analogue filter would try and remove. Now we oversample 16x to 705.6kHz and the max audio signal frequency with an input sample rate of 352.6kHz is 176.4kHz. Therefore the lowest beat frequency is 705.6kHz -176.4kHz = 529.2kHz.
AE: The separate and significant power supply: how does a digital source benefit from such extreme high quality power provision? Does it require high current as well as high quality? If not, wouldn’t a smaller but still high quality power supply suffice?
SS: Adding a separate high quality PSU has two benefits. One is removing the iron and copper toroidal transformer from the same box as the delicate DAC and analogue circuits. The transformer has stray magnetic fields and also mechanical vibration caused by magnetostriction. Secondly the separate low noise supplies prevent noise from one circuit section interfering with another circuit — for example the streamer module interfering with the DAC clock of I2V.
Looking further into detail, feeding elements such as the actual DAC itself with ultra clean power has great benefits. The DAC is sensitive to noise on its power supply. The DR regulator had a dramatic effect on the sound quality. We have four DR regulators dedicated to powering the DAC; these are placed directly under the DAC on a separate PCB. The DRs also feed the analogue circuits which are also sensitive to noise. None of the circuits require lots of power, it’s all about how clean they are. We like to use simple audio circuits and let them work in the best environment, clean of noises. As opposed to using more complex circuits with lots of feedback and simpler PSUs etc.
Analogue circuits have inputs that will be amplified other than the main audio input — from the PSU rails, magnetic and RF, physical vibrations and thermal… we work hard to minimise the other unwanted sources of noise.
AE: How sensitive to input voltage and ambient conditions are Naim components and power supplies? On set-up day we had 246V from the mains... and an ambient temperature something over 30C. Australia, eh!
SS: We design the input mains to tolerate -10% on 220V and +10% on 240V giving a usable range of 198V to 264V. Certainly as the mains voltage increases the internal heat dissipation increases and the transformer can get closer to saturation. We test new products in our environmental chamber that goes from -40 to +85. We do some little tricks inside to minimise temperature variation effects. One trick is to put sensitive components such as matched pairs of transistors under little thermal covers. These keep the temperature difference between parts more consistent as the outside temperature increases.
Some devices can improve their performance as they get warmer. Transistors can work better. When we talk about ‘warm up’ it can mean more than just physically warming up. One of the large effects of warm up is the chemical reaction in capacitors as their dielectric forms. This is the insulation part between the two plates of a capacitor.
We also analyse in CAD how circuits perform over a wide range of temperature. Here we use analysis called ‘Monte Carlo analysis’. An output from this may show the response of say a filter at different temperature; we can observe the variation and ensure it’s within tolerance at all temperatures.
AE: Why no XLR balanced outputs?
SS: We designed the ND 555 to work primarily with the NAC 552 pre-amplifier. The 552 has
single-ended inputs only.
The PCM1704 being a single-ended output DAC also means all circuits are single-ended. A balanced output in this case would mean an additional circuit to generate the ‘cold’ signal. We normally try to avoid additional circuits. Certainly with the 555 being typically sited close to the 552 there will be no problems with noise pick-up.
AE: This is the first Zigbee remote for Naim? Is this the beginning of a range-wide change?
SS: The ND555 remote is technically the second — the one shipped with Uniti is the first. However they are identical, except that the 555 remote has a meticulously machined aluminium outer and has green lights to reflect the materials used on the main units. We will use the ZigBee remote for all new products. It’s bi-directional and goes through walls.
Interview: Jez Ford. This interview accompanied our review of the ND 555 Network Player, which appeared in issue #1-2019 of Audio Esoterica magazine. The review can be read online by clicking here, and this interview should be read in the context of the review.