What the C10 Is, in Engineering Terms
Why distribute a clock at all?
In a multi box digital front end, each device derives its sample timing from a local oscillator. Where a digital interface carries clock information embedded in the data, as S/PDIF and AES/EBU do, the receiving device must recover that clock and lock to it, typically through a PLL. Recovery is imperfect, and the recovered clock inherits the character of both the source oscillator and the interface.
Distributing a common reference inverts the arrangement. Rather than each device timing itself and one device chasing another, all devices discipline their internal timing to a single external reference whose phase noise is lower than any of their internal oscillators. The benefit is not simply "more accurate frequency". Absolute frequency accuracy matters far less to audio than short term phase stability, which is why the specification that deserves attention is the phase noise curve rather than the ppm figure.
Why phase noise rather than jitter?
They describe the same physical phenomenon in different domains. Jitter is the time domain statement, phase noise the frequency domain one, and the two are related by integration across an offset range. Phase noise is the more informative specification because it shows where the instability lives. Close in noise, within a few hertz of the carrier, behaves differently from noise out at a kilohertz, and the audible consequences are not the same.
The C10 is specified at minus 115 dBc/Hz at 1 Hz offset, minus 129 at 10 Hz, minus 139 at 100 Hz and minus 145 at 1 kHz, with a floor at or below minus 170 dBc/Hz. The close in figure is the difficult one to achieve and the one that separates a serious reference from an ordinary oscillator.
The Oscillator
The oscillator in the C10 is EverSolo’s second generation clock design, the second iteration of their own low noise reference work rather than a first attempt at it.
Why an SC cut crystal?
Quartz resonators are cut from the crystal at specific angles, and the angle determines how the resonant frequency responds to temperature and to stress. The AT cut is singly rotated, cheap and adequate. The SC cut, meaning stress compensated, is doubly rotated. It exhibits a far flatter frequency against temperature characteristic near its turnover point and is substantially less sensitive to mechanical stress, including the stress induced by its own mounting and by thermal gradients during warm up.
SC cut crystals are harder to manufacture, must be driven at a specific overtone, and need more careful oven design. They are used where short term stability genuinely matters, which is exactly the case here.
What does the oven contribute?
An oven controlled oscillator holds the crystal above the highest expected ambient temperature at a point chosen so that the frequency against temperature curve is flat. The combination of an SC cut crystal and a well designed oven is what produces the quoted figures: under 0.003 ppm of drift across minus 20 C to plus 60 C, factory offset inside plus or minus 0.01 ppm, aging below 0.002 ppm per day, and short term stability better than 9E-13 at a one second gate time.
That last figure, Allan deviation at one second, is the one to look at if you care about short term behaviour, because it describes stability over the interval that matters to audio rather than over days.
Why does it need five minutes?
The oven has to reach its setpoint and the crystal has to come into thermal equilibrium with it, including the relief of stresses introduced by the temperature gradient during heating. Until that settles, the oscillator is still moving. The C10 reaches rated performance within five minutes at 25 C ambient and reports the process on its display, which is a practical courtesy rather than a specification.
Distribution
Why independent buffering matters more than output count
A clock distribution network is a transmission line problem. Each output must present a controlled 50 ohm source impedance into a 50 ohm cable terminated at 50 ohms, because an impedance discontinuity produces reflections, and reflections arriving back at the wrong moment degrade the edge quality of the signal that the receiving device is trying to use.
If several outputs are simply paralleled from one driver, the load presented by one device modulates what the others see, and any noise a connected device injects back onto its clock line propagates to its siblings. The C10 buffers each of its four BNC outputs independently, which isolates them from one another. The output is a square wave at 1.8 Vrms into 50 ohms, a level chosen so that the receiving device sees fast, clean edges.
Why offer frequencies other than 10 MHz?
Ten megahertz is the instrumentation convention, inherited from test and measurement, and a device fed 10 MHz must synthesise its own audio sample clocks from it, usually with a PLL. Every synthesis step is an opportunity to add phase noise.
The C10 also outputs 11.2896, 22.5792 and 45.1584 MHz, which are integer multiples of the 44.1 kHz family, and 12.288, 24.576 and 49.152 MHz for the 48 kHz family, plus 25 and 100 MHz. Where a receiving device can take a direct audio rate master clock rather than a 10 MHz reference, it can skip a synthesis stage. That is a meaningful architectural advantage, not a specification sheet flourish.
What is the 1PPS input actually for?
One pulse per second is the standard distribution format for disciplined time references, most commonly from GPS. Feeding the C10 a 1PPS reference lets its long term frequency be steered to that source, which addresses absolute accuracy and aging rather than short term phase noise. Every unit is calibrated for 1PPS timing at the factory. For audio purposes it is a completeness feature; the short term performance that matters to playback comes from the OCXO itself.
System Architecture: Where the C10 Does Real Work
The split source, T10 into DAC Z10
The strongest case is a two box digital source. A T10 streaming transport delivers digital to a DAC Z10, and in the ordinary arrangement the Z10 either locks to the clock embedded in the incoming stream or asynchronously reclocks it against its own oscillator. Either way, two oscillators are involved in one source, and the interface between them is where timing error is introduced.
Referencing both to the C10 collapses that to a single clock domain. The transport and the converter are no longer negotiating; they are both disciplined to the same reference. Two of four outputs are consumed, and the architecture of the source is simplified rather than merely improved.
The integrated source, DMP-A8 Gen 2 into AMP-F8
Here the analysis runs the other way, and it is worth being candid about it. A DMP-A8 Gen 2 contains the streaming, conversion and preamplification stages on shared internal clocking, and the AMP-F8 downstream is an analog power amplifier with no clock domain at all. There is no inter box digital interface to discipline, so the C10 can only improve the quality of the single reference already in use rather than eliminate a disagreement between two.
That is a smaller benefit, and it argues for spending elsewhere unless the system is heading toward separates. Introduce a DAC-R8 alongside the A8 and the calculus changes immediately, because you have recreated the two box digital chain where a shared reference does structural work.
Clocking the DAC-R8
The DAC-R8 accepts a 10 or 25 MHz external reference on a 50 ohm input and runs an internal 10 MHz OCXO with a PLL when no external reference is present. Driving it from the C10 substitutes a better reference at the top of that chain. In an R2R ladder converter the timing of the switching instants directly determines the fidelity of the reconstructed waveform, so the sensitivity to reference quality is real rather than theoretical.
An Honest Assessment of Returns
External clocking is one of the few areas of digital audio where the engineering case and the audible case can diverge, and it is worth saying plainly. The engineering case for collapsing two clock domains into one is sound. The size of the audible result depends on how good the internal oscillators already were, how the receiving device uses the external reference, and whether the rest of the system can resolve the difference.
In a split source of genuine quality, feeding a resolving system, the reported effect is improved image stability and a lower noise floor rather than any tonal change. In a single box digital front end the effect is smaller and sometimes not worth the outlay. This is a component for the end of an upgrade path, and we would rather you heard it in your own system before deciding.
Come and Hear It
We keep EverSolo set up and running in the showroom in Timonium, and a clock is exactly the kind of component that should be auditioned against the system it will join. Call or come by.