Timecode Drift: Why Your Audio and Picture Fall Out of Sync Over a Long Recording

The Documentary That Drifted Into Chaos

A documentary production shot a 90-minute uninterrupted performance recording -- a live concert -- using dual-system audio on a Sound Devices 788T and a camera that was jammed to timecode at the start of the evening. By the time the concert ended, the picture and audio timecodes matched at the start of the recording but were approximately 3.5 frames apart at the end.

The editor could sync the head, or sync the tail, but not both. Every song that was cut using tail sync started slightly off at the head. The production had assumed that jamming once at the start was sufficient. For a 90-minute uninterrupted recording on two separate crystal oscillators, it was not.

This post explains what timecode drift is, how to calculate how many frames of drift to expect from your equipment over a given recording duration, and the on-set practices that prevent this problem entirely.

The oscillator tolerance values referenced below are drawn from technical specifications published by Sound Devices, Zaxcom, Tentacle Sync, and Ambient Recording for professional production audio equipment.

What Crystal Oscillators Have to Do with Timecode

Timecode is generated by counting cycles from a crystal oscillator -- a quartz crystal vibrating at a precise frequency (typically 32.768kHz or a multiple of it). The camera counts these oscillations to track time. The audio recorder does the same independently. When you jam sync at the start of a recording, you align both counters at a single point in time. After that, each device runs on its own crystal.

No crystal is perfect. Every oscillator has a rated tolerance, expressed in parts per million (ppm). A crystal with a tolerance of ±2 ppm means it can run up to 2 microseconds fast or slow per second. This sounds trivially small. Accumulated over hours, it is not.

The drift formula is:

Drift (seconds) = ppm x 10^-6 x recording duration (seconds)

At ±2 ppm over 1 hour: 2 x 10^-6 x 3,600 = 0.0072 seconds = 0.17 frames at 24fps -- imperceptible.

At ±2 ppm over 4 hours: 2 x 10^-6 x 14,400 = 0.0288 seconds = 0.69 frames at 24fps -- approaching the edge of perceptibility on close cuts.

At ±2 ppm over 8 hours: 2 x 10^-6 x 28,800 = 0.0576 seconds = 1.38 frames at 24fps -- clearly audible on dialogue cuts.

At ±5 ppm (common in budget recorders) over 8 hours: 5 x 10^-6 x 28,800 = 0.144 seconds = 3.46 frames at 24fps -- a production-stopping problem.

At ±50 ppm (unrated consumer equipment) over 8 hours: 1.44 seconds = 34.6 frames -- completely unusable without manual sync correction on every cut.

Three Real-World Drift Scenarios

Example 1: Single-Day Narrative Production, Jam Sync at Top of Each Roll

A narrative feature production uses a Sound Devices MixPre-10M (rated ±2 ppm) recording dual-system to a Sony FX9 (rated ±5 ppm from factory, less with a jam). The production jams at the start of every roll -- roughly every 30 to 45 minutes. Maximum continuous drift between jams: 2 x 10^-6 x 2,700 = 0.0054 seconds = 0.13 frames. This is below any perceptible threshold. The per-roll jam strategy completely eliminates meaningful drift as a problem.

Example 2: Documentary Concert Recording, Single Jam at Start

The scenario described in the opening: 90-minute uninterrupted concert using Sound Devices 788T (±1 ppm) and a camera with a ±5 ppm oscillator after the jam warms up. Worst-case drift (camera at +5 ppm, audio at -1 ppm = 6 ppm relative drift): 6 x 10^-6 x 5,400 = 0.0324 seconds = 0.78 frames at 24fps. If the camera oscillator is genuinely at the ±5 ppm limit, drift over 90 minutes is under one frame -- annoying but solvable in post. If the camera's oscillator degrades during warm-up or runs hotter than spec, actual drift may exceed spec. The lesson: always assume worst-case ppm for the lowest-rated device in the chain.

Example 3: Multi-Day Reality Production, Recharged Recorders Without Jam

A reality production uses Deity TC-1 timecode boxes (rated ±0.2 ppm) on both camera and audio. The audio team recharges recorders overnight but forgets to re-jam after charging. The TC-1 maintains timecode through its internal battery during charging. If the timecode box was properly jammed the previous morning and has been running continuously (not power-cycled), drift accumulates from the original jam. At ±0.2 ppm over 30 hours: 0.2 x 10^-6 x 108,000 = 0.0216 seconds = 0.52 frames -- still manageable. But if the TC-1 was power-cycled during charging and re-started without a jam, it resets to an internal clock reference and drift accumulates from a new, unsynced baseline.

Crystal Oscillator Tolerance by Equipment Class

The table below shows published ppm specifications for common professional audio and timecode devices. Lower ppm ratings produce less drift over equivalent recording durations.

The Tentacle Sync E and Ambient ACC 501 represent the current standard for production timecode boxes in narrative and documentary work. At ±0.2 ppm, even an eight-hour continuous recording accumulates less than 0.2 frames of drift -- operationally irrelevant for any post workflow.

How to Manage Timecode Drift on Set: Step by Step

Step 1: Before the shoot, check the ppm specification for every device in your audio chain -- the camera body, the audio recorder, and any dedicated timecode generators. The weakest link (highest ppm) determines worst-case drift. Enter your recording duration and ppm values into the Timecode Calculator to see the expected maximum drift for your production day.

Step 2: Determine your jam sync schedule. For productions with natural breaks between takes (narrative, commercial), jam at the start of every roll -- roughly every 20 to 45 minutes. The accumulated drift per roll is negligible with any professional-grade recorder.

Step 3: For continuous or long-form recordings (concerts, events, observational documentary without breaks), establish a maximum drift budget and calculate the jam interval from the worst-case ppm. At ±5 ppm on the camera, a jam every 2 hours keeps drift below 0.036 seconds (under 1 frame at 24fps).

Step 4: Use a dedicated timecode generator (Tentacle Sync, Ambient Lockit) rather than relying on camera-to-recorder jam via BNC cable when possible. Dedicated TC boxes run at ±0.2 ppm and can be wirelessly distributed across all devices simultaneously.

Step 5: Log the timecode jam time and device IDs on the production sound report for every roll. This creates an auditable record for the post team. If sync issues arise in the edit, the sound report tells the editor exactly when each device was last jammed and enables the drift calculation to narrow down the source.

At the end of this process the post team has fully documented jam timestamps and knows the maximum drift exposure for any given clip.

Pro Tips and Common Mistakes

Pro Tip: On ARRI ALEXA cameras, the "Sync" LED on the timecode block goes green when the camera accepts an external LTC jam and amber when it is free-running. Monitor this LED during long takes. If the LED has gone amber, the camera has lost lock on the external timecode signal and is now running on its own crystal -- drift has been accumulating since the lock was lost.

Pro Tip: Tentacle Sync E boxes can maintain timecode via Bluetooth and sync to each other wirelessly between takes without a physical BNC connection. This means sound can jam all cameras in a multi-camera setup simultaneously from across the room, without stopping camera operations. For run-and-gun documentary and live events, this workflow replaces the physical cable-jam that is frequently skipped in fast-moving environments.

Pro Tip: When a sync issue is discovered in post, the fastest diagnostic is to check the audio waveform against the camera audio (if any exists) at both the head and tail of the affected clip. If the waveform aligns at head but drifts at tail, the cause is oscillator drift and the fix is a speed-correction retiming. If the waveform is offset by a fixed amount from head to tail, the cause is a jam that was applied incorrectly. These two problems have different solutions and are worth distinguishing before the editor tries to fix them manually.

Common Mistake: Jamming once at the start of the production day and not repeating. A single jam at the start of day is effective only if all devices run continuously and the camera's oscillator maintains spec. In practice, cameras are power-cycled between setups, recorders may lose power during transport, and oscillator performance varies with temperature. The fix: treat jam sync as a per-roll process on any production using dual-system audio.

Common Mistake: Not confirming sync after a lunch break or location move. During a location move, recorders and cameras are often powered off and back on. Each power cycle can introduce a timecode discontinuity -- the device resets to an internal reference rather than continuing from the jammed value. A quick re-jam after every power cycle costs 30 seconds and eliminates post-production sync issues entirely.

Frequently Asked Questions

What is the difference between LTC and MTC, and which one causes more drift?

LTC (Longitudinal Timecode) encodes timecode as an audio signal on a dedicated track -- a physical signal that the receiving device reads and locks to in real time. MTC (MIDI Timecode) encodes timecode as MIDI messages. Both can be used to distribute timecode, but LTC is the standard in film and broadcast production because it is hardware-level locking and not subject to the latency variation that MIDI data can introduce. Neither method eliminates drift on the transmitting device -- the receiving device locks to the transmitter's timecode, so both devices share the transmitter's ppm tolerance after jam.

How does timecode drift differ from audio sample rate drift?

They are related but separate problems. Timecode drift affects the timestamp associated with each clip -- the starting reference point used to align audio and picture. Audio sample rate drift affects the playback speed of the audio within a clip -- the rate at which samples are played back. A sample rate discrepancy makes audio appear to run slightly fast or slow relative to picture, which looks like drift but is caused by a different mechanism. For sample rate management in post, Sample Rates in Film and Audio Post covers the distinction and the relevant delivery standards.

Can timecode drift be corrected in post without manual clip-by-clip work?

Yes, if the drift is consistent and linear. A speed retiming tool (available in Premiere Pro, Avid Media Composer, and DaVinci Resolve) can stretch or compress an audio track by a calculated percentage to re-align it with picture from head to tail. The calculation is: correction factor = (actual clip length) / (intended clip length). If a 90-minute clip has 0.78 frames of drift, the audio needs to be stretched by approximately 0.036 percent. Most professional NLEs can apply this correction non-destructively to the entire clip.

What happens to timecode drift on a multi-camera shoot?

Drift is a per-device-pair problem. In a two-camera setup, Camera A may drift +1 frame over the day while Camera B drifts -0.5 frames -- producing 1.5 frames of inter-camera drift by end of day. For multi-camera productions, a dedicated timecode generator distributed via LTC to all cameras and recorders ensures that all devices share the same crystal reference after jam, dramatically reducing inter-device drift.

Related Tools

The Timecode Calculator converts between frames, seconds, and timecode values for any frame rate, and can help calculate the expected drift window for a given ppm tolerance and recording duration. For understanding the relationship between audio sync, sample rate, and timecode in your post workflow, Timecode in Film Production: The Complete Guide covers jam sync, LTC, drop-frame, and non-drop-frame with production examples.

For downstream audio delivery once the sync issue is resolved, Audio Delivery Standards for Film and Television covers the submission requirements for broadcast and streaming, and Sample Rates in Film and Audio Post covers the sample rate specifications that underpin the entire audio sync chain.

The 30-Second Habit That Prevents Hours of Post Work

Timecode drift is not a post-production problem -- it is a production problem with a post-production cost. The solution is a 30-second jam sync procedure on every roll of dual-system audio, combined with a dedicated timecode generator rated at ±0.2 ppm or better on any production with long continuous takes. The math is not complicated: at ±0.2 ppm, you can record for 8 hours continuously before accumulating even 0.2 frames of drift. At ±5 ppm without re-jamming, you can be 3.5 frames off after 8 hours. The choice between these outcomes is made in pre-production, not the edit suite.

This post covers single-camera and dual-system audio workflows. Multi-track live sound recording for concert films involves additional timecode distribution architecture that warrants separate treatment. What is your current jam sync interval -- per roll, per half-day, or per day -- and has it ever produced a sync issue in post?