{"id":9515,"date":"2026-05-03T12:05:06","date_gmt":"2026-05-03T10:05:06","guid":{"rendered":"https:\/\/felusch.de\/?p=9515"},"modified":"2026-05-20T10:21:58","modified_gmt":"2026-05-20T08:21:58","slug":"praezision-im-netzwerk-audio-over-ip-als-signalquelle-fuer-line-array-lautsprecher-systeme","status":"publish","type":"post","link":"https:\/\/felusch.de\/en\/praezision-im-netzwerk-audio-over-ip-als-signalquelle-fuer-line-array-lautsprecher-systeme\/","title":{"rendered":"Network Precision | Audio-over-IP as a Signal Source for Line Array Speaker Systems"},"content":{"rendered":"

Acoustic coherence in the field of tension between waveguide mechanics and network synchronization.
<\/em><\/h4>\n\n\n\n

v1.<\/em>1.1. | 2026-04-28<\/em> | \u00a9 Bodo Felusch <\/em><\/p>\n\n\n\n

Switch language<\/p>\n\n\n\n\n\n

\n \n\n<\/svg>\n\n \n
\n \n en <\/span>\n <\/div>\n \n \n
\n \n de <\/a>\n <\/div>\n\n \n\n<\/svg>\n<\/div>\n\n\n
<\/div>\n\n\n\n

<\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n

<\/p>\n\n\n\n


<\/p>\n\n\n\n

Reference: Timing Precision Requirements for Line Arrays<\/strong><\/h4>\n\n\n\n

The requirements for mechanical precision in line array loudspeakers are well-established. The waveguide is\nthe most precision-critical component, guiding the energy of the high-frequency drivers as coherently \u2014 i.e. as\nphase-linearly \u2014 as possible to the acoustic aperture. The race for the optimal waveguide has largely been run\nout, constrained by patent law. So why not open a new front?<\/p>\n\n\n\n

First, we need to establish a benchmark for the precision we are aiming for. Everything discussed below is\nevaluated at 20 kHz. We define the acceptable phase difference in the waveguide as 1\/4 of the wavelength \u2014\ni.e. a 90\u00b0 phase difference \u2014 yielding a summation result of +3 dB from two coherent sound sources at 0 dBr\neach. For mechanical construction this translates to an acceptable tolerance window of 4.3 mm across the full\noperating temperature range. Coupling between adjacent elements is excluded from this discussion; sufficient\nliterature already exists on that topic. <\/p>\n\n\n\n

An equivalent error budget at the upstream electronic level corresponds to approximately 12.5 \u00b5s of time\ndifference between two coherent signals at exactly 0 dBr. This is slightly more than one sample period at 96\nkHz, which has a duration of 10.4 \u00b5s. As a physical reference point: a temperature difference of just 1\u00b0C across\n10 m causes a propagation delay difference of approximately 51 \u00b5s \u2014 exceeding the entire electronic error\nbudget (~12.5 \u00b5s) by a factor of four. This strikingly illustrates that acoustic environmental conditions are often\nthe limiting factor in practice, not the electronics.<\/p>\n\n\n\n

New technologies such as beamforming \u2014 which aim to steer the directional behaviour of a line array segment\nelectronically \u2014 impose far more stringent precision requirements. This is almost certainly the reason why\nelectronically steered systems rely on fixed mechanical splay angles. When using FIR filters, the limiting factor\nat low frequencies is latency: at 96 kHz with 2048 taps, one must accept 10.6 ms of latency, with only 46.9 Hz\nof bandwidth per tap \u2014 all-pass filters can provide relief here. At high frequencies, the enclosure and drivers\nthemselves are the limiting constraint: to accurately control phase to within 180\u00b0 at 20 kHz, the physical\ntolerance must be kept below 8.5 mm. <\/p>\n\n\n\n

In short: at low frequencies, time is the limiting resource; at high frequencies, space is.<\/p>\n\n\n\n

Fundamentals of Synchronisation<\/strong><\/h4>\n\n\n\n

Clock Drift<\/strong>
Before mobile phones united us on a common time reference, free-running clocks defined our individual,\nabsolute, yet perpetually inaccurate sense of time. Mechanically constructed timepieces and quartz watches,\nfor instance, vary in their precision. Even two identical quartz watches will not remain in agreement over time.\nAssume a quartz oscillating at 32,768 Hz with a drift of \u00b110 ppm. If two such watches are both set to 00:00 on\nthe 1st of January and left running, they will show different times after just a month \u2014 they drift apart. In the\nworst case, the two clocks diverge by 1.7 seconds per day, and by 52 seconds after one month.\nAnyone who has recorded picture and sound on separate devices knows that both recorders must be\nsynchronised to maintain lip sync throughout the recording. The problem is exactly the same as with the two\nquartz watches: neither is locked to a common reference. Clock drift is a frequency offset \u2014 a mistuning of the\nrate at which the clock oscillator is supposed to run.<\/p>\n\n\n\n

<\/p>\n\n\n\n

Interface Jitter<\/strong>
When two digital audio devices are operated at the same sample rate, connected digitally in series, but each\nlocked to its own internal clock, the same problem recurs \u2014 they drift apart and audible clicks begin to appear.\nA Clock Leader must be defined, and all digitally connected devices must be synchronised to it as Clock\nFollowers. In this process, the clock signal is transmitted over a physical medium, giving rise to Interface Jitter\nbetween the digital interfaces. In serial (daisy-chain) cabling, this interface jitter accumulates and corrupts the\nclock signal. To put it concisely: the clock edges smear in time, and the receiver can no longer reliably\ndistinguish between a one and a zero. The underlying causes are cable-induced errors such as impedance\nmismatch, capacitance, reflections, excessive cable length, or simply physical damage and ageing. Star-topology\nwiring helps to keep interface jitter low. A clock signal corrupted by interface jitter can be regenerated through\nre-clocking: the signal is buffered and re-generated by an internal clock that is itself synchronised to the Clock\nLeader.<\/p>\n\n\n\n

<\/p>\n\n\n\n

Clock Wander & Clock Jitter<\/strong>
Beyond the fundamental long-term drift that every oscillator exhibits \u2014 no matter how precise \u2014 there are\nalso short-term fluctuations. Imagine listening to a motivating mixtape through headphones, running for 30\nminutes, fully in the groove. Then you sneeze, stumble, momentarily lose the beat \u2014 but a few seconds later\nyou are back in sync with the music. After four or five hours in the blazing summer heat, your muscles begin to\ntire and your step starts to drift off the beat. The same phenomenon occurs \u2014 with far smaller deviations \u2014 in\nevery clock source.\nBeyond drift, there are low-frequency fluctuations in phase \u2014 phase noise \u2014 caused by environmental\ninfluences such as temperature variation, power supply instability, or mechanical stress. This low-frequency\n\"wobble\" below 10 Hz is called Clock Wander. High-frequency phase noise above 10 Hz is called Clock Jitter,\nand it comes in two distinct varieties. Matter is fundamentally never at rest; the purity and properties of\nsemiconductors vary in quality. This gives rise to Random Jitter, which follows an unpredictable Gaussian\ndistribution \u2014 almost as if matter itself were an overarching clock source, the fundamental noise floor of time.\nDeterministic Jitter, by contrast, is consistent and repeatable \u2014 the consequence of our own design decisions.\nElectronic circuits suffer from crosstalk, power supply ripple, electromagnetic interference, and impedance\nmismatches. Clock jitter is a \"tremor\" around the target frequency at which the oscillator is supposed to run.<\/p>\n\n\n\n

<\/p>\n\n\n\n

Synchronisation Accuracy<\/strong>
Every digital transmission system requires a shared clock reference. The Clock Leader is the authoritative time\nreference for all Clock Followers, which lock to it using a PLL (Phase-Locked Loop) circuit that nudges their\ninternal oscillator faster or slower as required. How well a system achieves this is described by its\nsynchronisation accuracy. In Dante with PTPv1, the target synchronisation accuracy is 1 \u00b5s between a single\ntransmitter and receiver pair, and 2 \u00b5s in multi-receiver configurations. Under degraded network conditions, a\nguaranteed synchronisation accuracy of one sample length is maintained.<\/p>\n\n\n\n

What Time Is It \u2014 Really?<\/strong>
The most precise clock we know of is the atomic clock, which derives its time reference from the radiation\ntransitions of electrons in free atoms. Caesium, rubidium, hydrogen, and more recently strontium are the most\ncommon atoms used, with a drift of approximately one second per 300 million years. While this may seem like\nmore than enough precision to keep an appointment, it is in fact insufficient for the highly precise technology\nwe have built. Even an atomic clock drifts by 1 nanosecond after approximately 110 days.
If we want to synchronise to 1 ns precision globally \u2014 for example, to achieve 1-metre GPS positioning\naccuracy \u2014 we need a more elaborate approach. We therefore average the historical measurements of some\n400 atomic clocks at 60 institutions worldwide to form the International Atomic Time (TAI), from which the\nCoordinated Universal Time (UTC) is derived. This is, in essence, Leader Clock Number One. GPS satellites\ncarrying their own atomic clocks, synchronised to ground stations, form Leader Clock Number Two. These two\nsystems are not synchronised to each other; they run independently, typically diverging by no more than\napproximately 20 ns. UTC is not directly relevant to the rest of this discussion, but one fact is worth noting: due\nto the Earth's irregular rotation, leap seconds are periodically inserted to keep UTC aligned. GPS time, by\ncontrast, has been running continuously since 1980. GPS time therefore currently runs exactly 18 seconds\nahead of UTC. TAI had a 10-second head start over UTC at its launch in 1972, and a 19-second lead over GPS\ntime at the GPS epoch of 6 January 1980 at 00:00:00 \u2014 an offset that has remained constant since. TAI and GPS\nboth run continuously without leap seconds. As of May 2026, TAI leads UTC by 37 seconds (18 s + 19 s); this\nmust be taken into account when converting between PTP\/TAI and UTC.<\/p>\n\n\n\n

Reference: GPS Jitter<\/strong><\/h4>\n\n\n\n

The GPS system is an excellent source of high-precision, globally synchronised time information. The jitter level\ndepends heavily on the quality of the receiver: the receiver's internal oscillator must suppress the inherent GPS\njitter. Typical values range from a few nanoseconds to 50\u2013100 ns; at the higher end, the timing is effectively\nderived from the receiver's own oscillator and only disciplined by GPS.<\/p>\n\n\n\n

Reference: Wordclock-Jitter<\/strong><\/h4>\n\n\n\n

The datasheet of an industry-benchmark wordclock generator provides the following specifications:<\/p>\n\n\n\n