I85PREP
⌘ K
JC Music study materials
JC Music study resources

Clicks Every Three Seconds: Clocking, Buffers and the Overdub Session

JC Music · Put your knowledge to work in context.

All case studies

Clicks Every Three Seconds: Clocking, Buffers and the Overdub Session

A project studio records at 48 kHz, 24-bit through a USB interface that is also the clock master. An eight-channel ADAT preamp is connected for drum overdubs. During tracking, the drummer hears irregular clicks every two to three seconds, and the recorded files show brief dropout spikes aligned with them. The ADAT unit's front panel reads 'Internal 44.1 kHz' because it was last used with an older rig. The DAW buffer is set to 128 samples; the interface reports 11 ms round-trip latency, but the drummer says the click feels closer to 25 ms behind the kit. A limiter and a linear-phase EQ sit on the master bus. There is a paying client in the room, a five-hour session, and no spare converter or word-clock cable.

1. Rank the likely causes of the clicks and dropouts, and state the exact clock and sample-rate settings you would apply with the available hardware.

Read the worked answer

Ranking:

  1. Two clock masters. The ADAT preamp is running on its internal 44.1 kHz clock while the interface sends 48 kHz. Two unsynchronised clocks drift until samples are dropped or repeated - the classic signature is a click every few seconds, not continuous noise, and the interval scales with the drift rate rather than with the audio content.
  2. Sample-rate mismatch on the optical link. Even with clocking corrected, a 44.1 kHz ADAT stream into a 48 kHz session will either be sample-rate-converted (if the interface supports it) or fail to lock.
  3. Cable or connector fault - check after the clock issue is resolved, not before.

Settings with the hardware on hand:

  • ADAT preamp: clock source = ADAT/external, sample rate = 48 kHz. It must slave to the interface, since the interface is the USB device the DAW is locked to.
  • Interface: clock source = internal, sample rate 48 kHz, and confirm the ADAT input shows 'locked' at 48 kHz before arming any track.
  • Verify by watching the lock indicator for a full minute with the drummer playing; clicks that persist after lock indicate a cable or optical-port problem rather than clocking.

If the ADAT cannot be switched to external clock, the alternative is to make it the master and slave the interface - but only one device may be master.

2. The drummer reports roughly 25 ms of perceived latency at a 128-sample buffer. Reconcile that with the reported 11 ms and state what you would change for tracking.

Read the worked answer

The arithmetic: at 48 kHz one sample is about 0.0208 ms. A 128-sample buffer is therefore 2.67 ms per buffer; a round trip of two buffers is about 5.3 ms. The interface's reported 11 ms already includes A/D and D/A converter latency plus driver overhead, so 11 ms is a plausible hardware round trip.

The extra 14 ms is almost certainly plugin latency on the master bus. A linear-phase EQ works by processing a block of samples and must delay the output to keep the phase response symmetrical; depending on the plug-in and its FFT size that can be tens of milliseconds. A look-ahead limiter adds its own delay. If the drummer is monitoring through the DAW, that latency is added on top of the interface figure.

Actions for tracking:

  • Bypass or remove the linear-phase EQ and the limiter from the master bus; if a limiter is needed for the drummer's comfort, insert it after the monitor path or use a zero-latency monitor mixer.
  • Use the interface's direct monitoring for the kit and the DAW's click, and confirm the click offset compensation if the DAW offers it.
  • Keep the buffer at 64–128 samples for tracking (1.33–2.67 ms per buffer). Raise it to 512–1024 samples (10.7–21.3 ms) only for mixing, where latency is inaudible.

Re-measure by having the drummer play to the click and comparing transients against the grid before committing to a five-hour session.

3. Should the master-bus chain stay active while tracking? Justify your answer in terms of workflow and technical risk.

Read the worked answer

No. Three reasons:

  1. Latency. As above, linear-phase processing and look-ahead limiting add delay to the monitor path. The musician plays to what they hear, so a delayed monitor mix produces a performance that sits late against the grid.
  2. CPU stability. A heavy master chain on every track, including record-armed ones, raises the chance of buffer underruns. Dropouts that occur while tracking are unrecoverable - the take is spoiled, and the client is paying for the time.
  3. Judgement and irreversibility. Mixing decisions made under tracking pressure are made without balances, and a limiter can disguise a clipping converter or a poorly set gain stage. Any processing committed on the way in cannot be undone.

Workflow to propose: track with the master chain bypassed and a conservative gain structure (peaks around −12 to −6 dBFS), using direct monitoring for the performer. Save the chain as a template and re-enable it for the mix, where the buffer can be raised and latency no longer matters. If the client insists on hearing 'the sound,' build a separate monitor-bus chain that feeds only the headphones - not the record path - and document it in the session notes.