Jitter

The title of this article is ambiguous. Further meanings are listed under Jitter (disambiguation) .

As jitter [ dʒɪtɚ ] ( Engl. For, fluctuation 'or' variation ') is defined as the time clock jitter in the transmission of digital signals, a slight variation of accuracy in the transfer clock (ger .: Clock ). Jitter is normally undesirable as an interference signal. More generally, jitter in transmission technology is an abrupt and undesirable change in signal characteristics. This can affect amplitude as well as frequency and phase position. The jitter is the first derivative of a delay (engl .: Delay ). The spectral representation of the temporal deviations is called phase noise . Jitter is not to be confused with quantization errors .

In network technology, jitter is also used to describe the variance in the transit time of data packets. This effect is particularly disruptive in multimedia applications on the Internet (such as Internet radio and Internet telephony ), since it means that packets can arrive too late or too early to be able to be output in good time. The effect is reduced by a so-called jitter buffer, a special "data buffer", but at the price of additional runtime , which is particularly annoying in dialog applications. This effect also plays a role in process control technology . Critical process information must be sent and received within a certain time. If the jitter becomes too great, the timely arrival of the critical process information is not guaranteed.

rating

Measurements of jitter

Various methods are available for evaluating jitter in the form of measured values. In the field of digital signal processing such as digital audio in the context of AES-3 or in the case of digital video signals in the context of the Serial Digital Interface (SDI), the temporal jitter is expressed as a relative quantity in Unit Interval (UI). A UI corresponds to the duration for a symbol. In the case of a binary transmission, this is the time for the transmission of a bit, as shown in the figure below for an exemplary binary sequence “01001”. The transitions between two successive different bits are shown stylized in light blue by a smoothed signal edge . The jitter causes the actual signal curve over time to deviate from the ideal signal curve, which is shown as a dark blue line, in the area of ​​the signal edge.

A large jitter causes increased symbol crosstalk and, as a result, an increased bit error rate, which can also be represented by a reduction in the horizontal opening in the eye diagram . The deviation from the ideal point in time of the signal edge can also be expressed as an absolute time specification in addition to the specification relating to the symbol rate in UI. Usual absolute values, shown as A _j or also as peak-to-peak in the diagram, have a few 100 fs (femtoseconds) to a few 100 ps (picoseconds) in serial digital transmissions in the megabit to gigabit range. With slower transmissions, absolute jitter times down to the microsecond range may be permissible, depending on the method.

The deviations, the spectral representation of which is referred to as phase noise , are divided into periodic or deterministic and random jitter components. The periodic components can be described with a fundamental oscillation, denoted in the diagram with the duration T _j , since it is the source of the greatest temporal deflection. It is superimposed on higher spectral components with lower amplitudes and random jitter, which vary in strength depending on the cause.

The jitter frequency f _{j of} the fundamental is given as:

${\displaystyle f_{j}={\frac {1}{T_{j}}}}$

For the reception of digital data streams and the timing of the sampling times, a clock recovery is necessary on the receiver side , which uses phase-locked loops in various forms, among other things . Those control loops can directly compensate for slow-moving, i.e. low-frequency, spectral components of the jitter by readjusting the local oscillator , while higher-frequency jitter components are suppressed by the low-pass behavior of the loop filters and can thus lead to sampling errors.

For numerical evaluation, it is therefore necessary to subdivide the jitter into its spectral components and evaluate them separately from one another or, depending on the transmission method, define permissible limit values ​​for the individual frequency ranges. For example, the spectral jitter components that are less than 10 Hz are generally referred to as wander .

The names of the higher-level jitter components are not chosen uniformly in the literature or for the individual transmission methods. For example, in the context of digital video transmissions (SDI) with timing jitter, those spectral components between 10 Hz and 1 kHz (with SD-SDI, in the standard SMPTE 259M) or between 10 Hz and 100 kHz (HD-SDI, in the standard SMPTE 292M). As a rule, these jitter components can still be compensated directly by the control loops. Spectral components above this are referred to as alignment jitter , as they can lead directly to sampling errors and are not compensated for by the phase-locked loops.

species

Density function of deterministic jitter (DJ) and random jitter (RJ) as a function of the unit interval (UI)

Jitter is divided into deterministic jitter (DJ) and random jitter ( English random jitter, RJ ). In transmission systems, both components appear superimposed with different weightings. In contrast to random jitter, deterministic jitter does not have a normal distribution , is always limited in amplitude and is described by its maximum peak-to-peak deviations. It can be determined, among other things, by corresponding symbol sequences and is divided into the following parts:

• periodic jitter components: The cause is typically external periodic interference signals which couple into the transmission system.
• data-dependent jitter components: these components depend on the transmitted data sequences and are caused by symbol crosstalk .
• Jitter component due to unequal pulse widths ( English duty cycle jitter ) is caused by different slew rates in the rising or falling signal edge.

A distinction must be made between these proportions of deterministic jitter and random jitter, which has a normal distribution and is caused, among other things, by thermal noise , uneven doping of the impurities in the semiconductor material used, and other random disturbances such as cosmic radiation . Random jitter is described by its standard deviation .

n	BER
6.4	10 -10
6.7	10 −11
7th	10 −12
7.3	10 −13
7.6	10 -14

Total jitter

The total or total jitter TJ is a combination of deterministic jitter (DJ) and random jitter (RJ) in the form:

${\displaystyle TJ=DJ_{\mathrm {p-p} }+2\cdot n\cdot RJ_{\mathrm {rms} }}$

The weighting factor is determined by the permissible bit error rate (BER). Common values ​​for the bit error rate, such as with Ethernet , are values ​​such as 10 ⁻¹² with . Further values ​​are summarized in the adjacent table. ${\displaystyle n}$ ${\displaystyle n=7}$

Digital audio systems

Another example of jitter are errors that can occur when converting analog signals to digital signals . In the scanning ( English Sampling ), a fixed period is used, for example 22.67 microseconds in the range of the respective audio signals at 44.1 kHz amplitude values are read out.

Even with digital audio systems according to the AES3 standard, jitter is determined by the spectral distribution in

• low frequency interface jitter and
• higher frequency sampling jitter

divided. Sampling jitter occurs in digital audio systems, among other things, in analog-to-digital converters , asynchronous sampling rate converters and digital-to-analogue converters .

Mathematical Definitions

Periodic jitter:

${\displaystyle J_{\mathrm {per} }=T_{\mathrm {per} }(1)-T_{0}}$

This is the period of the first oscillation after the trigger event and the ideal period. ${\displaystyle T_{\mathrm {per} }(1)}$ ${\displaystyle T_{0}}$

Cycle-to-cycle jitter:

${\displaystyle J_{\mathrm {cc} }=\max(T_{\mathrm {per} }(n)-T_{\mathrm {per} }(n+1))}$

The maximum deviation from one period to the next is determined.

Accumulated jitter

${\displaystyle J_{\mathrm {ac} }(n)=T_{\mathrm {per} }(n)-n\cdot T_{0}}$

With accumulated jitter, starting with a trigger event (e.g. a rising edge of a clock signal), the jitter is related to this event. The longer the clocks are in the future, the greater the shift will be if the jitter is not evenly distributed.

literature

• Dennis Derickson, Marcus Müller: Digital communications test and measurement. High-speed physical layer characterization . Prentice Hall, Upper Saddle River NJ 2007, ISBN 978-0-13-220910-6
• Julian Dunn: Jitter. Specification and Assessment in Digital Audio Equipment . 1992 ( PDF )
• Dan Lavry: On jitter . 1997 ( PDF )
• Wolfgang Maichen: Digital Timing Measurements. From Scopes and Probes to Timing and Jitter . (= Frontiers in Electronic Testing; 33). Springer US, Berlin 2006, ISBN 0-387-31418-0
• Johann Christoph Scheytt: Clock and data recovery circuits with automatic selection of the bit rate for bit rate flexible optical transmission systems . Dissertation, Ruhr University Bochum 2000 ( PDF )
• Mike Story: Timing Errors and Jitter . 1998 ( PDF )
• John Watkinson: The Art of Digital Audio . 3. Edition. FocalPress, Oxford et al. a. 2001, ISBN 0-240-51587-0
• Understanding and Characterizing Timing Jitter , Tektronix Primer, Sept. 2012 ( PDF ).
• Digital Timing: Clock Signals, Jitter, Hystereisis, and Eye Diagrams , National Instruments Tutorial, Dec. 30, 2016 ( PDF )

Web links

Commons : Jitter  - collection of images, videos and audio files

Individual evidence

1. ↑ ^{a b} Jitter Measurement for Serial Digital Video Signals . Tektronix Inc. company publication, 2006 ( [1] [PDF]).
2. ↑ ITU-T Recommendation G.810: Definitions and Terminology for Synchronization Networks, August 1996
3. ↑ A Guide to Understanding and Characterizing Timing Jitter . 55W-16146-1. Tektronix Inc. company publication, 2003 ( PDF ).
4. ^ Jitter Working Group Technical Report, T11: Methodologies for Jitter Specification. InterNational Committee for Information Technology Standards (INCITS), 1998, available online: PDF, 678 kB ( Memento from July 31, 2016 in the Internet Archive ).