Introduction

A spread spectrum quartz crystal clock oscillator is an oscillator that has the output frequency intentionally modulated in order to reduce the electro-magnetic interference (EMI) on the output signal. Spread spectrum quartz crystal clock oscillators are best used in applications that require a reduction of EMI emissions in order to pass the FCC EMI regulations. These include copiers, FAX machines, modems, set-top boxes, scanners, printers, LCD displays, computers graphic cards, hard drives, interface controllers, and PCI/CPU/Memory buses.

Spread spectrum crystal oscillators, like those produced by Ecliptek Corporation, reduce the EMI at the clock source, rather than at locations later down in the clock stream. By reducing the EMI at the clock source, supplemental shielding enclosures and/or filtering components may not be required, reducing overall system costs and improving overall EMI performance.

By modulating the output signal, the EMI on the output signal is 'spread' over a larger frequency 'spectrum'. The total amount of energy is still present, but the spreading of the output power over the frequency band results in a reduction of EMI at any one frequency. Regulatory bodies like the FCC have maximum limits for peak EMI emissions (emissions at any one frequency within the spectrum). Thus, a quartz crystal clock oscillator can be used to pass FCC regulatory EMI test requirements by reducing EMI peak emissions.

A programmable spread spectrum crystal oscillator consists of a fundamental mode quartz crystal controlled oscillator and a flash programmable integrated circuit. Programming of the oscillators is typically performed by the oscillator supplier and offers improved delivery times as compared with non-programmable oscillators. The memory within the device contains programmable functions that control the operating characteristics of the device. These include the output frequency, modulation frequency, output frequency spread spectrum percentage, power down/tri-state function, and duty cycle.

A block diagram of the Ecliptek EPS13D2 series programmable crystal clock oscillator is shown in Figure 1. At the heart of the oscillator is a programmable high resolution Phase Locked Loop (PLL). The PLL consists of a reference counter divider, a feedback counter multiplier, spread spectrum modulation signal, and an output divider. Utilizing a programmable spread spectrum IC and the proper programming methodology, spread spectrum crystal oscillators can be programmed with specific values that define the operating characteristics of the oscillator.

Figure 1: PLL Block Diagram

Electrical Specifications

Spread spectrum programmable quartz crystal oscillators typically have output frequencies in the range of 1.000MHz to 170.000MHz and power supply voltages of 3.3VDC or 5.0VDC. Due to the oscillator CMOS design structure, device input current is typically less than 20mA (at VDD = 3.3Vdc). The output is designed to provide a low voltage HCMOS (LVHCMOS) output topology so as to optimize circuit load matching and signal performance. Output load termination is normally specified as 15pF maximum with rise and fall times of less than 3nS maximum (measured at 20% to 80% of the waveform) and duty cycle of 50% ± 5% (measured at 50% of waveform). Through the use of a fundamental mode bulk acoustic wave (BAW) quartz crystal, improved overall frequency temperature stability can be achieved. Typical industry standard offerings are ±100ppm, ±50ppm, and ±25ppm stabilities over the operating temperature ranges of 0°C to 70°C (commercial) or -40°C to +85°C (Industrial).

For proper system operation, output clock jitter performance is critical. Period jitter is the measure in the time domain and is specified in picoseconds (pS). Peak to Peak and rms period jitter are commonly specified for non-modulated clock oscillators and are defined over a number of cycles. Due to the output frequency modulation, only cycle-to-cycle period jitter is specified on spread spectrum clock oscillators. Cycle-to-Cycle period jitter is measured in the time domain and is defined as the worst-case clock period deviation of adjacent cycles. Because the modulation frequency is substantially slower than the output frequency, the spread spectrum modulation has very little impact upon the cycle-to-cycle or short-term jitter. Cycle to cycle period jitter specifications can vary greatly depending upon frequency, input voltage, and modulation width and profile.

Spread spectrum clock oscillator products are offered in two different tri-state output logic control function options to facilitate the customer's use of in-process assembly testing or for the use of multiple clocks. These oscillators contain flash programmable power down (PD) and tri- state/output enable (TS) options for power management. If the power down function is selected, all active circuitry within the oscillator is shut down when the voltage at the control pin is set to a logic low state. In this condition, the output signal is three-stated (tri- state) and the input current on the power supply line is negligible (stand-by current specification of 50µA).

If the tri-state/output enable option is selected, the output is three-stated (tri- state) when the voltage at the control pin is set to a logic low state. In this condition, the oscillator and PLL continue to operate. However, the output signal is now high impedance and the input current on the power supply line is only slightly decreased (approximately 60% of nominal I_CC) from normal operating current.

Packaging and Construction

Spread spectrum crystal clock oscillators are typically available in industry standard packages. These include the 14 pin DIP (dual in-line), 8 Pin DIP, 5mm x 7mm x 1.8mm ceramic SMD four pad, and 5mm x 7mm x 1.8mm ceramic SMD six pad SMD packages. These products consist of a single IC and a fundamental mode bulk acoustic wave (BAW) quartz crystal packaged inside a hermetically sealed package. The DIP oscillators are manufactured with a metal resistance welded package. The package has a sealed metal cover that is case grounded for improved EMI performance. The SMD oscillators are manufactured with a ceramic leadless chip carrier that has gold plated contact I/O pads are typically RoHS compliant (Pb-free).

These SMD oscillators have either four I/O pads or six I/O pads and are seam sealed with a metal cover that is case grounded for improved EMI performance. The pads for the four I/O package are defined as Power (VDD), Ground, Output, and Power Down/Output Enable. The additional pads on the six pad ceramic SMD package are a no connect pad and a spread spectrum enable/disable output control pad. The additional control pad is an input pad designed to be toggled between a logic high and a logic low state. This control pad is not to be confused with the power-down or enable/disable tri-state function pad.

When the SS enable/disable control pad of the six-pad ceramic SMD is placed in a logic low state (or no connect), the output signal of the oscillator is clocking with the spread spectrum feature enabled (output frequency is modulated). When the device is placed in a logic high state, the output signal of the oscillator is clocking with the spread spectrum feature disabled (output frequency is not modulated). A timing diagram of the spread spectrum (SS) enable/disable function for the Ecliptek EPS13D3 series programmable crystal clock oscillator is shown in Figure 2.

Figure 2: Timing Diagram

Electro-magnetic Interference

EMI is short for Electro-magnetic Interference and is defined as a naturally occurring phenomenon when the electro-magnetic field of one device disrupts, impedes, or degrades the electro-magnetic field of another device by coming into proximity with it. EMI can cause two or more electronic devices to interfere with each other and affect their performance and operation. For example, when you are using your cordless phone or laptop computer, you do not want it to interfere with your television reception. Frequency sources such as quartz crystal oscillators, phase lock loop (PLL) synthesizers, and other types of clock signal generation schemes are a major source of EMI in electronic circuits. Therefore, EMI reduction is a major concern for designers of electronic products utilizing these clock schemes. Conventional methods of EMI reduction include multiple ground and power planes, discrete component filtering, and enclosure shielding. These methods are commonly practiced and can have substantial cost impact to the overall product. An alternative to some of these EMI reduction techniques is the implementation of a programmable spread spectrum clock oscillator. The use of such an oscillator can significantly improve EMI and reduce overall system cost.

EMI is a measurement of radiated energy from a frequency source and is typically measured in dBmV/m (decibel-volts per meter) at a given frequency. This parameter is larger for higher amounts of radiated energy. Thus, the more energy emitted from a frequency source, the larger the resulting electric field and EMI. When defining the frequency spectrum, one wants to distinguish between the peak electro-magnetic emissions and the average electro-magnetic emissions. The average emission is defined as the average dBmV/m level over a given frequency spectrum of an oscillator output signal. Peak emission is defined as the maximum dBmV/m level at any frequency over a given frequency spectrum of an oscillator output signal.

Federal Communications Commission (FCC)

Electro-magnetic Interference (EMI) is subject to very strict regulations by the US Federal Communications Commission (FCC) and other international regulatory bodies whose goal are to limit the amount of EMI electronic devices emit and to prevent damage to the human body and interference between electronic devices. The FCC's Class A regulations apply to industrial applications and the Class B regulations apply to residential or consumer applications. Computer and peripheral hardware applications typically are concerned with compliance to Class B regulations. Today, FCC regulations are primarily concerned with peak emissions at any given frequency, not the average emissions over a given frequency spectrum. Thus, a circuit designer should focus their EMI design efforts with reducing the peak emissions at any given frequency within the frequency spectrum, not the overall average emissions within the spectrum. Figure 3 shows a FCC Class B plot of power (dBµV/m) versus frequency (MHz) for the peak emission requirements (at 10 meters).

Figure 3: FCC Class B Peak Emissions

EMI Reduction

In order to understand the importance of a spread spectrum oscillator, a thorough understanding of frequency modulation is required. Figure 4 shows a plot of output amplitude (power) versus frequency for a modulated and un-modulated center-spread spectrum clock oscillator. This figure illustrates the significance between the frequency span and the amplitude for a given modulated and un-modulated spread spectrum clock oscillator. By modulating the output frequency over a frequency spectrum, a reduction in output amplitude can be achieved. This reduction in output amplitude correlates with a reduction in radiated energy (EMI).

Figure 4: EMI Reduction Plot

There are two major factors that significantly affect the amount of peak EMI reduction for a spread spectrum clock oscillator: Output Frequency Modulation Width and Frequency Modulation Profile.

Figure 5 shows a plot of output frequency versus time for an output of a linear (triangular) modulated spread spectrum clock oscillator. As you can see from the figure, the output frequency has a minimum (F_MIN), center (F_C), and maximum (F_MAX) frequency. The output frequency is swept linearly though a range of frequencies rather than being held at one constant frequency. This 'range' parameter is often called output modulation width, output frequency spectrum, or frequency spread percentage. The maximum and minimum output frequencies are often stated as a percentage (%) with respect to the center frequency. Programmable crystal clock oscillators offer a range of programmable output frequency modulation widths from 0.5% to 5.0% (F_MAX minus F_MIN). The wider the modulation frequency spread percentage, the larger the bandwidth of frequencies over which the energy is distributed, and therefore the more EMI peak reduction.

Figure 5: Output Frequency Modulation Width

Modulation Profiles

There are three major types of frequency modulation profiles: Sinewave, Linear (or triangle), and Non-linear (or optimized). Each of these three modulation profiles result in different EMI reduction performance. An example of each modulation profile and the resultant output is shown in Figure 6.

Figure 6: Frequency Modulation Profile and
Resultant Frequency Spectra

The sinewave modulation profile is not typically used in spread spectrum clocks due to its large edge peaking. The linear modulation scheme has improved peaking at the edges, but the small trough in the middle of the profile often reduces the total EMI reduction. The best modulation scheme is the non-linear or optimized modulation profile. This profile reduces edge peaking and troughing, producing the most efficient EMI reduction profile. An example of a linear profile is shown in Figure 5 and of an optimized profile in Figure 7.

Figure 7: Non-Linear Frequency Modulation Profile

Since the frequency modulation width is fixed and independent of the frequency modulation profile, the total radiated EMI is spread over the frequency modulation width. The goal of a spread spectrum oscillator is to spread the EMI energy evenly over the frequency modulation width, so as to eliminate any peaks or troughs. Using a sinewave or triangular profile can result in increases in peak EMI emissions. Using an optimized modulation profile evenly distributes this EMI energy over the frequency modulation width. As shown in Figure 7, the output modulation frequency (Fm), often called sweep rate, is defined as the inverse of the modulation period. For the subject spread spectrum quartz crystal clock oscillators, the output frequency modulation is in the range of 30kHz to 60kHz.

Center and Down Spread Modulation

Programmable crystal clock oscillators are offered with two standard output frequency modulation options; 'Center Spread' and 'Down Spread'. Figure 8 shows an example of these two options.

Figure 8: Center and Down Spread Options

The instantaneous output center frequency (F_C) is approximately the midpoint of the minimum frequency and the maximum frequency. The instantaneous output frequency will therefore always vary between F_MIN and F_MAX. The instantaneous minimum (F_MIN) and maximum (F_MAX) output frequencies are stated as a percentage (%) with respect to the center frequency. In Figure 8, the center spread diagram provides an example of a device with a ±1.0% center spread percentage. In this example, if F_O were 100MHz, typical frequencies for F_MIN, F_C and F_MAX would be 99MHz, 100MHz, and 101MHz, respectively.

When a system cannot tolerate an operating frequency higher than the nominal frequency (often called over-clocking), then a down spread option should be considered. In Figure 8, the down spread diagram provides an example of a device with a ±2.0% down spread percentage. For this example, if a customer was concerned about over-clocking and had a maximum operating frequency requirement of 100MHz (F_O), typical frequencies for F_MIN and F_MAX would be 98MHz and 100MHz, respectively.

Utilizing a proper design and programming methodology, spread spectrum quartz crystal clock oscillators can achieve significant reductions (up to 20dB at the 7th harmonic) in output EMI emissions. The output frequency, spread percentage, modulation profile, and the measurement harmonic frequency are the major factors that determine the EMI reduction.

Spread spectrum crystal oscillators reduce the EMI at the clock source, rather than at locations later down in the clock stream. By reducing the EMI at the clock source, supplemental shielding enclosures and/or filtering components may not be required, reducing overall system costs and improving overall EMI performance.