대한민국 - 플래그 대한민국

통화 선택을 확인하십시오:

한국 원화
인코텀즈:FCA(배송점)
관세, 통관료 및 세금은 인도 시 징수됩니다.
통화에 따라 ₩60,000 (KRW) 이상 주문 시 대부분 무료 배송
신용카드로만 결제 가능합니다

미국 달러
인코텀즈:FCA(배송점)
관세, 통관료 및 세금은 인도 시 징수됩니다.
통화에 따라 $50 (USD) 이상 주문 시 대부분 무료 배송

Bench Talk for Design Engineers

Bench Talk

rss

Bench Talk for Design Engineers | The Official Blog of Mouser Electronics


Timing is Everything:  A Look at Oscillators, Clocks, Buffers and Redrivers Mike Parks

In the space it takes us to blink an eye (300 to 400 milliseconds is the average) a modern embedded core can perform millions of operations. It’s almost unfathomable for us to perceive such tiny fractions of time, especially since the evolution of timekeeping began with sundials and water clocks, with the first (fairly) accurate clock arriving in 1656. Modern digital technology operates very comfortably at these breakneck speeds because of decades of advancements in timing technologies such as crystal oscillators and clock generators.

 

Timing is a crucial aspect of digital systems, ensuring data smoothly progresses through processor pipelines and interconnected systems that can talk with each other.  Many simple, low speed applications may simply rely on a microcontroller’s internal RC oscillator to provide the necessary clock signal. However, internal oscillators may be too slow and noisy for some designs, or you may need multiple devices to share a single clock. Whatever the case, there are instances where an external timing solution is necessary.

 

High speed telecommunications equipment and bus interfaces are highly susceptible to timing problems. Among the problems facing clock timing design:

 

  • Skew: Problems arise when the pulses of the clock signal arrive to different components at different times.
  • Jitter: Clock signals need steep rise and fall transitions for the system to get optimal performance. Deviations from the ideal clock transition adds to clock uncertainty.
  • Power: Clock circuits switch constantly, thus the clock is a major factor of a device's power consumption.
  • Noise: Clocks tend to run at RF frequencies, thus extra care must be considered when routing timing traces.

 

Crystal Oscillators

 

A crystal oscillator, also referred to as a crystal resonator, is the simplest external timing solution to implement. The timing device is built using a piezoelectric material, commonly quartz, sandwiched between two metal plates. The crystal translates the mechanical resonance into an electrical signal with a set frequency. You can purchase crystal oscillators that produce pulses that range in frequency from from a few kilohertz to hundreds of megahertz. A key advantage of crystals is their very large Q-factor, which promotes very good frequency stability.

 

Crystals are two terminal devices and they rely on a microcontroller to provide some internal driver circuitry to get the crystal to generate the clock signal.  In addition, two external load capacitors are needed, attached between each pin and ground to provide stability.  Check the datasheet of your microcontroller for recommended values for the capacitors based on the intended operating frequency. Typical oscillator modes for externally driven microcontrollers include low power/low speed (LP), average speeds (XT), and high speed (HS) operation.

 

 

 

 

Figure 1: Crystal Oscillators such as this 24MHz crystal from Abracon let you add external timing quite easily to your project.

 

 

Clock Generators

 

A more complete solution is a clock generator, also called a frequency generator. These integrated circuits offer a lot of functionality for a moderate increase in cost. Some of these advantages include the ability to generate multiple frequencies that can also be changed dynamically by the microcontroller using the I2C or SPI serial communication protocols. Clock generators offer significant benefits when used to generate multiple frequencies. Smaller footprint and component counts lead to better reliability as well as lower costs.

 

On the negative side, crosstalk is one issue that must be considered if you use a clock generator to create multiple clock frequencies. Crosstalk occurs when a clock signal induces a phantom signal on another trace. Good board layout is key to fighting crosstalk.

 

Figure 2: Though more complicated to integrate, clock generator ICs offer far greater functionality, like this TI LMK61e2 oscillator.   Image from LMK61e2 datasheet by Texas Instruments.

 

 

Buffers and Redrivers

 

When a single clock source has to drive multiple devices, it becomes necessary to compensate for fanout limitations. Fanout is a reality of gate driven design. While in theory it is often acceptable to model transistor gates as having “infinite” input impedance, the reality is not so straightforward. Given the fact that the gates do leak, it is only possible for an output to drive a certain number of inputs before the combined leaks take a toll on the signal. A fanout buffer allows for a single source to drive the clock inputs on multiple devices by giving the input signal a boost and duplicating it on multiple output lines.

 

Redrivers (also called repeaters) are necessary when signals have to travel significant distances. They can be used for both onboard PCB interconnections as well as connecting two distinct physical devices. Some examples include a PCIe bus onboard a motherboard or the USB connection between the motherboard and an external hard drive. Redrivers can help to compensate for long distance channel losses at the transmitter and clean up the signal at the receiver.  Long distance may be defined in inches or centimeters for onboard trace purposes, or several feet/meters for external cable extensions.

 

Figure 3: Redrivers, such as this PTN36221A USB SuperSpeed Redriver from NXP, allow signals to travel farther and improves performance. Image from PTN36221A datasheet by NXP.

 

It All Comes Down to Communication

 

Without robust and reliable timing components, modern digital life would go silent. Clean clock signals allow devices to communicate reliably by providing a common drumbeat. Without those rapidly oscillating pulses to provide a timing reference, devices would not be able to properly read the data coming in on the data lines. Imagine an assembly line where no two workstations could communicate with each other. If even one station got behind, the rest of the factory floor would soon be in disarray. So, too, would modern telecommunication and high speed peripherals become unusable if their timing was not able to remain reliably synchronized. Timing really is everything.

 

 



« Back


Michael Parks, P.E. is the co-founder of Green Shoe Garage, a custom electronics design studio and embedded security research firm located in Western Maryland. He produces the Gears of Resistance Podcast to help raise public awareness of technical and scientific matters. Michael is also a licensed Professional Engineer in the state of Maryland and holds a Master’s degree in systems engineering from Johns Hopkins University.


All Authors

Mostra altri Mostra altri
Visualizza i blog per Data

Archivio