Clock signal
The use of a clock signal (also just clock; English clock signal or clock ) is a method to ensure the correct timing when operating an electronic circuit. In particular, many digital circuits require a corresponding signal for the temporal coordination or synchronization of the actions of several circuits (in particular that of flip-flops ) within complex digital systems ( switchgear ). In addition, the frequency of the clock signal can serve as a reference frequency; it thus ensures that electronic watches, for example, run smoothly. Certain analog circuits, for example switched capacitor filters , also require an exact clock frequency.
It is usually a periodic signal that is characterized by its frequency (called clock frequency or rate ) or its reciprocal value ( period duration ). It alternates between two logic levels , marked in the adjacent sketch with H for high and L for low . A prominent example of a clock signal is the system clock ( system clock ) of the operating speed of many components, in particular in a computer, microprocessor determined. An example of an aperiodic clock signal is clocked data transmission , for example with the SPI or I²C interfaces .
Demarcation
In technical jargon, the term “ clock ” is often used , whereby it must be inferred from the context whether this is a clock signal, clock frequency or clock cycle .
General
When a periodic clock signal is present, it is generated by an oscillator such as a crystal oscillator . A common oscillator circuit for generating clock signals is the Pierce circuit . Further characteristic properties of a clock signal (in addition to the frequency or period duration) are the accuracy ( manufacturing tolerance , temperature dependency , phase noise and jitter of the clock signal source ) , the amplitude ( voltage ) and the duty cycle . The application results in related requirements that affect the selection or the design principle of the oscillator (for example ring oscillator with very low accuracy requirements or atomic clock with highest accuracy requirements ).
Ideally, a clock signal is a square wave signal . In fact, however, the waveform of a clock signal is often more sinusoidal. Often the clock frequency is almost as high as the technological maximum frequency of the switching mechanism. In this case, a waveform with a low proportion of harmonics , which also potentially causes less interference , is sufficient to satisfy the requirement for the edge steepness .
Clock generation
In order to generate a square-wave clock, in addition to the quartz oscillators and the control circuit, quartz oscillators are now used in electronic circuits. The advantages of these components are the low tolerance of the generated frequency and the high stability of the frequency over the permissible temperature range, the aging resistance of the component and the permissible range for the operating voltage of the components.
Modern processors and their auxiliary components on the motherboard of a computer require several different clock signals, since the CPU , for example, runs at a much higher clock frequency than external interfaces. Within the CPU, too, frequencies are switched dynamically depending on the operating situation, primarily to save energy. A master oscillator is usually responsible for providing such diverse clock signals, which derives all the required frequencies from a crystal frequency via its own frequency divider or phase-locked loop (PLL) .
CPUs and GPUs are offered in versions for certain clock frequency ranges for which they were specified during development and tested during manufacture. Since reserves are planned in these areas due to operational safety and unavoidable production tolerances, many users - especially from the computer game area - strive to push the limits. Many processors offer special control registers for this overclocking , which can be used to set a division factor - the multiplier - that specifies how the operating frequency should be divided from the clock frequency. Such settings are risky, however, since in extreme cases data can be lost or the processor can be destroyed. Therefore, there are companies that offer such overclocked computers or graphics cards as part of their range of services that they have previously tested for adequate operational reliability.
Clock signal in integrated circuits
The circuits that use the clock signal for synchronization can, depending on the design, become active either during the rising or falling signal edge (with Double Data Rate (DDR) both edges are used); one speaks of edge control or edge triggering . The clock signal is often referred to as CLK in data sheets and diagrams . By means of clock gating , the clock signal can be selectively switched off in integrated circuits for certain circuit parts that are not currently required in order to reduce the average power consumption.
Most more complex integrated circuits require a clock signal to synchronize different parts of the chips and to compensate for gate delays. As these components are becoming more and more complex and also tend to be faster due to technical progress, the delivery of an accurate and uniform clock signal to all circuits is becoming an ever greater challenge for chip developers. The prime example of such complex chips are microprocessors , the central components of modern computers.
Working speed of processors
The clock frequency of a processor indicates the frequency with which the processing units are clocked.
It is given in Hertz (Hz). Since the frequency can be several billion Hertz, the numbers are often abbreviated with the help of prefixes such as Giga (G) for billions or Mega (M) for millions (in embedded systems , however, kHz are sometimes also common). For example, a processor clock frequency of 1 GHz means a clock period of one nanosecond .
With current processors, the clock frequency corresponds to the frequency with which machine cycles can begin, with older processors this could also be significantly lower than the clock frequency. E.g .:
- Intel 8051 : clock frequency 24 MHz, a new machine cycle can begin every 12 clock cycles → effective cycle frequency 2 MHz / 500 ns
- Texas Instruments TMS320C40: clock frequency 50 MHz, a new machine cycle can start every 2 clock cycles → effective cycle frequency 25 MHz / 40 ns
The effective processing speed cannot be definitively deduced from this either, because on some processors even simple instructions still take 10 to 20 cycles, on others several complex instructions can be started each cycle. What is done in 4 cycles on one CPU takes a hundred thousand cycles on another.
Multi-core processors allow separate clock frequencies for each core as well as for globally used resources (L3 cache, PCI-Express , RAM interface, QPI ); these can continue to change according to current requirements.
The data throughput of a processor results from its clock frequency and the data transfer rate of its connection to the main memory . The computing power (measured in MIPS or FLOPS, for example ) is not only dependent on the frequency, but also on the overall architecture of the processor. Even with processors that have the same instruction set use, serious computing power differences can indicate the cause of, for example, in the IPC rate (IPC by itself at the same clock rate English instructions per cycle , instructions per clock cycle ), company-specific characteristics (for example, SIMD Extensions ) or can be due to the memory bandwidth already mentioned. The IPC rate indicates how many instructions per clock cycle a processor can process simultaneously through parallelization . The processor with a higher IPC rate therefore creates more arithmetic operations per clock cycle and therefore calculates faster.
literature
- Vojin G. Oklobdzija, Vladimir M. Stojanovic, Dejan M. Markovic, Nikola M. Nedovic: Digital System Clocking . High-performance and low-power aspects. Wiley-IEEE Press, 2003, ISBN 978-0-471-27447-6 .
Individual evidence
- ↑ Clock. In: ITWissen.info - technology knowledge online. DATACOM Buchverlag GmbH, accessed on September 29, 2017 .
- ^ Paul Horowitz , Winfield Hill: The Art of Electronics . 2nd ed. Cambridge University Press , Cambridge, United Kingdom 1989, ISBN 0-521-37095-7 , pp. 282 .