Power Supply Filtering for Clock Sources, Repeaters, and PLL Circuits

If your oscillator has poor power supply immunity or if it must work inside a noisy system, give it some extra power supply filtering. The amount of filtering required depends on how much a reduction in jitter you must achieve. Determining a precise value for required jitter reduction is almost impossible because all the parameters vary:

• Jitter performance is not specified on many clock sources. When your purchasing department buys a different brand of oscillator, the jitter will change.
• Noise in a system changes when different brands of integrated circuits (perhaps faster switching ones) are assembled .

Nevertheless, you need to do something. Most manufacturers of clock sources recommend a circuit somewhat like the one in Figure 12.60, often without the resistor. Using the values and parasitic assumptions listed in Table 12.5 and the layout in Figure 12.60, this filter achieves better than to 20 dB of power supply noise reduction in the frequency band from 10 MHz to 1 GHz (10 9 Hz). Cascading two sections roughly doubles the attenuation. The purpose of the resistor is to preclude resonance at the frequency f C where the filter begins to function. That frequency f C , called the low-pass cutoff, is given by

Equation 12.19

 

Figure 12.60. This filter provides reasonable attenuation of power supply noise from 10 MHz to 1000 MHz.

Without the resistor R1 installed, the cutoff-frequency resonance appears in Figure 12.61 at a frequency of 500 KHz. A reasonable value of R 1 that provides the right amount of damping to prevent such a resonance at f C is

Equation 12.20

 

Figure 12.61. The frequency response of a realistic filter is an amalgamation of compromises.

With this resistance installed, the loss slope of the filter is “20 dB per decade. One decade above f C you can expect an attenuation of roughly 20 dB.

Above 100 MHz the parasitic shunt capacitance of the inductor ( C L,SHUNT ) begins to short out the inductor. At the same frequency the parasitic series inductance of the capacitor ( L C,SERIES ) substantially degrades its performance. This is a general example of the principle that at high frequencies, components do the opposite of what you want (inductors turn into capacitors and vice versa). At 1 GHz the parasitic effects render the filter totally ineffective .

The frequency at which a one-stage filter becomes totally degraded depends on the values of the parasitic elements. It is usually somewhere in the vicinity of this value:

Equation 12.21

 

where	C L,SHUNT is the parasitic shunt capacitance of the inductor, F, and
	L C,SERIES is the parasitic series inductance of the network of capacitors, H.

For the component values listed in Table 12.5 this resonance should appear at 1 GHz. As expected, a debilitating resonance appears at that frequency in Figure 12.61. The solution to this problem is to include a short trace on the downstream side of inductor L 1 . Provided the trace is laid out with its input and output well separated, it exhibits almost no end-to-end shunt capacitance. This trick increases the series impedance of the overall L 1 structure, disrupting the parasitic resonance at 1 GHz.

Table 12.5. Parasitic Models for Components in Power Filter

	Resistance	Inductance	Capacitance	Circuit model
L 1	.1	1 uH	100 pF
R 2	2.2	1 nH	1 pF
C 3 , C 4	.1	0.5 nH	.047 uF

Filters designed for wideband operation are built as a cascade of multiple sections, each section scaled to provide coverage in successively higher frequency bands, which is what has been done here. Before you attempt any type of advanced filter design, it is crucial that you measure and understand the parasitic parameters of your components. For inductors, the main parasitic elements are a shunt capacitance and a series resistance; for capacitors the main parasitic elements are a series inductance and a series resistance. These values are an integral part of any filter you build and must be included in your calculations.

When laying out any high-frequency filter, take care to keep the input and output well separated. All capacitors should preferably connect directly to a solid reference plane with large vias, at least 500 m m (20 mil) diameter. Keep all extraneous circuit traces as short as possible (less than 2.5 mm, or 0.1 in.), unless included intentionally as part of the circuit. Surface-mounted components, in the smallest size practicable, work best.

POINT TO REMEMBER

• Filters designed for wideband operation are built from a cascade of multiple sections, each section scaled to provide coverage in successively higher frequency bands.

 

12.12.1 Healthy Power

Article first published in EDN Magazine , March 30, 2000

Hurtling toward the airport in late-afternoon traffic, huddled in the backseat of a taxi, Ernie's fingers flew across the keys of his laptop as he scrambled to put the finishing touches on his latest digital-system design. He finalized the netlist and typed "send" just as the taxi turned off the freeway and into the airport. Perfect timing! His job was done. Thirty minutes later he stepped onto the Jetway a free man, bound for Aspen, Colorado.

Ernie had meticulously researched every aspect of his design, especially the power system. Every bypass capacitor was the proper size, and every dielectric layer was the proper thickness . Ernie relaxed in the first-class cabin , sipping champagne ”a model of supreme confidence.

You need not use a fancy high-impedance probe to measure power supply noise.

 

Or maybe not. Maybe Ernie didn't have time to research the power system. Maybe he just copied the same old approach he had used last year. Maybe the design took longer than he had planned, and in the crunch for time at the end, he had to give up his vacation. Does this situation sound familiar?

However you design your board, take the time to check the health of its power system when the board comes back from fabrication. Use an oscilloscope to directly measure the noise present on V CC (with respect to ground). Because the natural impedance between power and ground is low (less than 1 W ), you need not use a fancy high-impedance probe to measure power supply noise. You can use a plain coaxial cable directly soldered to the V CC and ground nodes of the pcb. Connect the other end of the coax to your scope and engage the scope's internal 50- W terminator.

To connect the coaxial cable to your board, remove one bypass capacitor and solder the coax signal and ground connections directly to the capacitor's V CC and ground mounting pads (Figure 12.62). This arrangement keeps the exposed coaxial signal conductor short, so it doesn't pick up much extraneous noise. Make the measurement at several locations. At high frequencies, the power system is a distributed circuit, so noise may vary across the board.

Figure 12.62. Use the mounting pads for a bypass capacitor to directly measure V CC noise with a coaxial probe.

Your noise will be a small AC signal superimposed on a relatively large DC bias equal to your power supply voltage. If you can't adjust the vertical position on your scope far enough to clearly see the noise, you may improve your resolution by using your scope's AC-coupling mode. The mode has a hitch, however: Although many scopes permit AC coupling, and many scopes provide an internal, built-in 50- W termination, you often cannot simultaneously engage these features. If this situation exists with your scope, you may want to provide an external AC-coupling circuit. This circuit consists of a DC-blocking capacitor in series with the coaxial connection to the board.

You can use a bypass capacitor already present on the board to perform the DC-blocking function. Lift off one of the capacitors, lay down some insulating tape on its ground pad, and re-solder the V CC end of the capacitor. The previously grounded end of the capacitor now sits on the insulator and is not connected to anything. Connect your coaxial signal connection to the previously grounded end of the capacitor and connect your coaxial ground connection to any nearby ground via. A 0.1- m F capacitance feeding into the 50- W load of your oscilloscope produces a high-pass filter with a time constant of 5 m sec, which comfortably passes all frequencies greater than 100 KHz.

Alternatively, you may build an external DC-blocking black-box attachment for your scope. The black box should contain one series-connected, 0.1- m F capacitor. It should have SMA coaxial-cable connectors attached to both ends of the capacitor with 50- W traces. A well- constructed black box should easily produce good performance through about 1 GHz.

Checking your power system's health always returns useful information. If you see too much noise, you know you have some serious work cut out for you. If you see very little noise, you may have the opportunity to save some money, space, and weight on your pcb by stripping out some of the bypass capacitors or reducing their sizes. Either way, measuring the noise on the power system gives you useful information that will help improve your design.

POINT TO REMEMBER

• Observing the noise between V CC and ground always returns useful information.

 

12.12.2 Clean Power

Article first published in EDN Magazine , August 3, 2000

Figure 12.63 illustrates the typical setup used to provide so-called quiet power for a sensitive analog circuit. Good applications for this LC-filter structure include oscillators , PLLs, and fiber- optic receivers. This filter reduces the differential noise that the analog component X between terminals AVCC and DGND perceives.

Figure 12.63. Noise from DGND flows through C2, making DGND and AVCC the same at high frequencies, thus eliminating the differential noise V DIFF across circuit X.

What about the absolute noise on AVCC? Compared with a true center-of-the-Earth ground-reference point, does the filter reduce the absolute noise on AVCC? Careful consideration of this question may lead you to a better understanding of the purpose of power supply filtering.

First, consider the matter of a 0-V potential reference. First-year electrical-engineering texts normally teach this concept in conjunction with the study of Kirchoff's laws, which form the basis of all modern electrical engineering. According to this method, every circuit contains one 0-V reference node. The equations then define all other voltages in terms of their potential differences from the reference node. Kirchoff's equations are so generally useful and so widely taught that engineers rarely stop to question their applicability.

Unfortunately, the problems of ground noise, electromagnetic radiation, and ESD susceptibility do not succumb to Kirchoff's analysis. These problems violate one of Kirchoff's first and most important assumptions: that all the electromagnetic fields in a circuit must be well-contained within compact, discrete circuit elements. When electromagnetic fields ravage the territory between circuit elements, the zero-potential concept evaporates.

For example, when measuring the potential difference between two points on the ground plane of a high-speed digital processor card, you must connect wires (or probes) to these two points and then feed the wires over to the inputs of your measuring equipment. Already, you have a problem. As any EMI professional will tell you, the space surrounding any processor card is filled with intense , high-frequency electromagnetic fields. These fields interact with your wires, inducing noise. The induced noise shows up in your measurement, and there's no way to eliminate it. Even worse , when you move the wires, the noise changes. The measurement and the measurement technique influence each other. Just as in relativistic physics, you, the observer, become part of the circuit.

A simple low-pass filter does not eliminate noise on AVCC. It merely makes AVCC and DGND the same .

 

With electromagnetic noise present, you can talk sensibly about potential differences only between points that are collocated ”that is, points so close that the total field strength between those points is negligible. Global 0-V reference potentials do not exist within large, high-speed digital systems.

Lacking a good global 0-V reference, then, does it make sense to talk about reducing the noise on AVCC? Yes, provided that you are interested only in reducing the differential noise between AVCC and DGND in the local vicinity of X, a job that the circuit in Figure 12.63 admirably performs .

To see how this circuit works, assume that at operational frequencies, the impedance of L1 is much greater than the impedance of X, and the impedance of C2 is much less than that of X. Component L1 thus operates as an open circuit, and node AVCC is shorted more or less directly through C2 to DGND. Given unavoidable high-frequency noise on DGND, C2 serves to inject that same noise directly onto AVCC, ensuring that AVCC and DGND perfectly track each other at high frequencies.

This filter does not eliminate noise on AVCC. It merely makes AVCC and DGND the same , reducing the differential noise between the two. Power supply filters always work that way. They copy junk from one circuit onto another so that the two match. For circuits such as oscillators, PLLs, and fiber-optic receivers, which don't reference other external grounds, noise matching between AVCC and DGND is generally all you need.

Circuits such as A/D converters that reference two ground systems may impose additional constraints. When working with an A/D converter, you need to connect all the relevant grounds together while ensuring that no high-speed currents can flow through the attachment point. The absence of high-speed currents eliminates local magnetic fields, ensuring the applicability of Kirchoff's laws near the attachment point. All the grounded metal near your A/D converter then truly rests at the same potential, and the circuit works.

Engineers often talk about "cleaning up" the AVCC supply. Power supply filters don't do that. If you want to clean up your AVCC plane, use soap. If you want to minimize the differential noise between AVCC and DGND, use a filter.

POINT TO REMEMBER

• A power-supply filter does not eliminate noise ”it merely copies junk from one circuit node to another, eliminating the difference between them.