The variation of a signal from its nominal timing value, resulting in a slightly different received signal from the ideal signal, is known as jitter.
Jitter does not cause a variation in the amplitude of the signal. The variation due to the jitter is in the signal phase, width and period. It is an unwanted factor that can lead to an increase in the bit error rate of a serial link. Due to the design simplification and high-performance advantages, high-speed serial buses are in use in high-performance designs. Data is transferred from one point to another within the system using serial data links. A clock and data recovery circuitry are used to accurately send the data in the system from sender to the receiver.
The accurate interpretation of the data at the receiving side mainly depends upon having the accurate information of clock edges. Many factors can affect the clock (timing signal), including electromagnetic interference, vibration and temperature, among others. There can be different factors at different locations that can affect the clock signal and introduce jitter because transmitting and receiving devices can be located anywhere in the world. If jitter is present in the clock signal, then there will be an increased chance of misinterpreting the receiving data. This could result in receiving a corrupted data file, low-quality audio or video file or other data errors, rendering the information received useless.
Each application has its own tolerance for jitter. However, a jitter of a single nanosecond can lead to a reduction in the bit resolution. Ideally, the time between two samples during analog to digital converting should be the same, but if jitter is present, then the time between two samples varies. There are different types of jitter, and the effect of jitter on the signal depends upon the type of the jitter.
Types of jitter
Random jitter is unbounded and can take any shape — that is why it is also known as Gaussian jitter. It can be caused by a random source at a random interval, thus making it unpredictable jitter. The main source of random jitter is the thermal noise of an electrical component. Short noise and flickers can also cause random jitter.
Deterministic is a predictable jitter and has a bounded peak to peak value. Deterministic jitter is further sub-categorized as period jitter, bounded uncorrelated jitter and data-dependent jitter.
Causes of device jitter
The high-speed performance demand of a small, compact device, where all components are integrated on a single chip, can affect signal integrity due to analog interference. PCB traces, conductors, vias and nearby components can all affect the signal integrity and introduce jitter. In high-performance, large scale integrated electronics, jitter is one of the major design considerations. Jitter is generated mainly from the interconnects.
- Absolute jitter changes the clock signal to a different position from where it was expected to be, and it can be measured with the help of a network analyzer.
- Period jitter is also known as cycle jitter, and it is a variation in a single cycle of the clock signal. It can be measured with the help of an oscilloscope.
- Cycle-to-cycle jitter is a variation in two adjacent cycles of the clock signal, and it can also be measured with the help of an oscilloscope.
How to measure and reduce jitter
The main measuring instrument is an oscilloscope that is used as reference equipment to test the device under inspection. Period jitter is measured by randomly calculating the clock duration of one period for 1,000 times minimum. After this, the peak to peak values, mean and standard deviation are calculated by using the recorded data. The same procedure is used to calculate the cycle to cycle jitter. It is measured by randomly calculating clock duration of two periods for 1,000 times minimum. The only difference is that in period jitter, one period is used to make 1,000 samples, while in cycle to cycle jitter, two periods are used to make samples, and then the difference between two of them is calculated. The mean and standard deviation is then calculated by using the recorded data.
It is possible to reduce the level of jitter in a signal by using an anti-jitter circuit. Many digital-to-analog converter (DAC) and analog-to-digital converter (ADC) systems use these circuits for clock and data recovery. The general principle of their work is that they tend to align output pulses close to the idealized curve. These circuits can only minimize the jitter factor, and it is not possible to eliminate it. A phase-locked loop is an example of this circuit.
Effect of jitter
Clock signals are important in embedded systems. Jitter can cause a delay in these signals that can severely damage the performance of the system. Embedded applications where synchronization is needed are sensitive, and a high jitter can lead to system failure.
Transmission control protocol (TCP)-based applications are most sensitive and are badly affected. High jitter can cause TCP-based applications to work poorly. These applications will still work, but jitter will damage the performance.
Jitter is not acceptable in real-time applications. For example, high jitter in video conferencing or voice over internet protocol (VoIP) will fail because two-way communication is not possible. A jitter of 100 mS or higher can fail these applications.
Jitter is typically an unwanted factor that can reduce the performance of the system. Yet, there is an application regarding this: self-generated jitters are fed to the electronic devices to analyze their performance. This testing could help determine the minimum tolerable jitter that a device or system can afford without compromising the performance.
