Jitter and Clocking Explained

Jitter occupies an odd place in audio discussion: it is invoked constantly, defined rarely, and used to justify a great deal of expenditure. The concept itself is simple and worth understanding properly, because it is the one respect in which a digital connection is not merely a data pipe. It is also, in well-designed modern equipment, largely a solved problem — which makes knowing what actually addresses it more valuable than knowing what claims to.

Timing, Not Data

Start with the distinction that clears up most confusion. Digital audio involves two separate things: which numbers arrive, and when each number is converted back to a voltage.

The numbers are robust. Digital transmission has enormous noise margins, and in practice the data arrives bit-perfect or fails audibly — a dropout, a click, silence. There is no gradual degradation of the values themselves.

The timing is a different matter. A digital-to-analogue converter reconstructs a waveform by producing each sample's value at a precisely spaced moment, at exactly the sample rate. If those moments are not evenly spaced — if the clock that triggers them wobbles — the reconstructed waveform is distorted, even though every number was correct. That timing uncertainty is jitter, and it is measured in picoseconds.

Where Jitter Comes From

Three sources account for nearly all of it.

Clock recovery. S/PDIF, AES/EBU and HDMI embed timing in the data stream, so the DAC has to extract it — usually with a phase-locked loop that follows the incoming transitions. Anything that shifts those transitions in time shifts the recovered clock: cable reflections from an impedance mismatch, slow rise times, the transmitting device's own timing. Because the transitions depend on the data being sent, the resulting jitter is correlated with the audio, which is the worst kind.

The oscillator itself. Every clock has phase noise. A basic crystal oscillator is adequate; a temperature-compensated or oven-controlled one is better. This contribution is random rather than correlated, and in a competent DAC it is already far below audibility.

The power supply and the board. Noise on the clock's supply rail modulates its output, and high-frequency digital activity elsewhere on the board couples in. This is why layout, supply regulation and isolation often matter more to real-world jitter than the specification of the oscillator itself.

How Modern Designs Solve It

The engineering answer, arrived at over the last two decades, is to stop trying to follow the source's timing at all.

Asynchronous USB inverts the relationship: the DAC's own local clock governs conversion, and the DAC signals the computer to send more data when its buffer needs filling. The source's timing quality becomes irrelevant, because nothing downstream is trying to track it. This is the single most effective structural fix available, and it is why asynchronous USB is usually a DAC's best-performing input.

Buffering and reclocking achieve much the same for stream-clocked inputs. Incoming data is written into a memory buffer using the recovered clock and read out using the DAC's own clock, decoupling the two. Provided the buffer is deep enough to absorb the difference, arrival jitter is discarded rather than passed on.

Asynchronous sample rate conversion takes a more aggressive approach, resampling the incoming stream onto the DAC's own clock domain. It is extremely effective against jitter, at the cost of a mathematical resampling step that purists dislike — though in practice modern implementations are transparent.

Galvanic isolation on the input prevents electrical noise from the source reaching the clock circuitry in the first place, which addresses the supply-noise contribution rather than the timing one.

External Clocks: When They Help and When They Do Not

Some DACs accept an external word clock, and dedicated master clocks are sold at serious prices. The logic sounds compelling — a better clock ought to be better — but the reality depends entirely on the architecture.

In a studio, an external master clock exists to keep multiple devices in sync so that digital audio can pass between them without sample slips. That is a synchronisation requirement, not a quality one, and it is entirely valid.

In a domestic playback system, feeding a DAC an external clock usually means the DAC must now slave to an incoming clock signal — travelling down a cable, through a receiver circuit — instead of using the oscillator sitting millimetres from its conversion stage. Unless the DAC was specifically architected to be clocked externally, this can easily make matters worse rather than better. The general rule: the best place for a clock is as close to the converter as possible, and a well-designed DAC with a good internal clock has little to gain.

What Actually Deserves Your Attention

If jitter is a concern in your system, these are the interventions that have real effect, in order.

• Use asynchronous USB where the DAC supports it, preferably with galvanic isolation on that input.
• Choose a DAC with buffering or reclocking if you will be using S/PDIF, coaxial or optical inputs.
• Use correctly specified cables — true 75 Ω coax, 110 Ω for AES/EBU. Impedance mismatches cause reflections, which is a genuine jitter mechanism.
• Remove the computer from the chain with a dedicated streamer, which eliminates a large source of electrical noise as well as timing variability.
• Ignore the picosecond figure in isolation. A number without a spectrum tells you nothing about audibility, and a modern DAC's jitter is already well below the threshold.

The broader point is worth stating: jitter was a real and audible problem in the 1990s, when clock recovery was primitive and buffering was rare. The architecture that fixed it — local clock authority, deep buffers, isolated inputs — is now standard in even modestly priced equipment. Understanding it is useful mainly so you can recognise which claims about it still hold.

Quick Reference: Jitter at a Glance

Aspect	Detail
What it is	Timing uncertainty at the moment of conversion, measured in picoseconds
What it is not	Data corruption — the numbers themselves arrive intact
Audible effect	Modulation sidebands around signal content, not added hiss
Worst kind	Correlated jitter, produced by recovering a clock from the audio stream
Main sources	Clock recovery, oscillator phase noise, power-supply and board noise
Best structural fix	Asynchronous USB, or buffering and reclocking on stream-clocked inputs
External master clock	Essential in studios for sync; rarely an improvement for home playback
Current status	Well below audibility in competent modern DACs

More Guides